diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4032 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.989094874591058,
+  "eval_steps": 500,
+  "global_step": 570,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.008724100327153763,
+      "grad_norm": 6.035512972360855,
+      "learning_rate": 1.4035087719298246e-06,
+      "loss": 1.1029,
+      "step": 1
+    },
+    {
+      "epoch": 0.017448200654307525,
+      "grad_norm": 6.042841036030158,
+      "learning_rate": 2.8070175438596493e-06,
+      "loss": 1.1024,
+      "step": 2
+    },
+    {
+      "epoch": 0.026172300981461286,
+      "grad_norm": 5.910306635602092,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 1.0986,
+      "step": 3
+    },
+    {
+      "epoch": 0.03489640130861505,
+      "grad_norm": 4.536951510086868,
+      "learning_rate": 5.6140350877192985e-06,
+      "loss": 1.0538,
+      "step": 4
+    },
+    {
+      "epoch": 0.04362050163576881,
+      "grad_norm": 2.3773466328483686,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 0.9941,
+      "step": 5
+    },
+    {
+      "epoch": 0.05234460196292257,
+      "grad_norm": 1.999261332126332,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 0.999,
+      "step": 6
+    },
+    {
+      "epoch": 0.061068702290076333,
+      "grad_norm": 4.369629342726399,
+      "learning_rate": 9.824561403508772e-06,
+      "loss": 0.9889,
+      "step": 7
+    },
+    {
+      "epoch": 0.0697928026172301,
+      "grad_norm": 4.899527942424933,
+      "learning_rate": 1.1228070175438597e-05,
+      "loss": 0.9631,
+      "step": 8
+    },
+    {
+      "epoch": 0.07851690294438386,
+      "grad_norm": 5.748165018609449,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.9524,
+      "step": 9
+    },
+    {
+      "epoch": 0.08724100327153762,
+      "grad_norm": 4.763126150981281,
+      "learning_rate": 1.4035087719298246e-05,
+      "loss": 0.952,
+      "step": 10
+    },
+    {
+      "epoch": 0.09596510359869138,
+      "grad_norm": 3.167454561583767,
+      "learning_rate": 1.543859649122807e-05,
+      "loss": 0.9095,
+      "step": 11
+    },
+    {
+      "epoch": 0.10468920392584515,
+      "grad_norm": 3.3676041736516433,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.89,
+      "step": 12
+    },
+    {
+      "epoch": 0.1134133042529989,
+      "grad_norm": 2.8110117483164676,
+      "learning_rate": 1.824561403508772e-05,
+      "loss": 0.8538,
+      "step": 13
+    },
+    {
+      "epoch": 0.12213740458015267,
+      "grad_norm": 2.0519859086879286,
+      "learning_rate": 1.9649122807017544e-05,
+      "loss": 0.85,
+      "step": 14
+    },
+    {
+      "epoch": 0.13086150490730644,
+      "grad_norm": 1.8373157533967603,
+      "learning_rate": 2.105263157894737e-05,
+      "loss": 0.8349,
+      "step": 15
+    },
+    {
+      "epoch": 0.1395856052344602,
+      "grad_norm": 1.6156997409278058,
+      "learning_rate": 2.2456140350877194e-05,
+      "loss": 0.829,
+      "step": 16
+    },
+    {
+      "epoch": 0.14830970556161396,
+      "grad_norm": 1.2400567097625959,
+      "learning_rate": 2.385964912280702e-05,
+      "loss": 0.8108,
+      "step": 17
+    },
+    {
+      "epoch": 0.15703380588876772,
+      "grad_norm": 1.1069155525681142,
+      "learning_rate": 2.526315789473684e-05,
+      "loss": 0.8029,
+      "step": 18
+    },
+    {
+      "epoch": 0.16575790621592149,
+      "grad_norm": 1.0769626887248518,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.7858,
+      "step": 19
+    },
+    {
+      "epoch": 0.17448200654307525,
+      "grad_norm": 0.7016586049675165,
+      "learning_rate": 2.8070175438596492e-05,
+      "loss": 0.7816,
+      "step": 20
+    },
+    {
+      "epoch": 0.183206106870229,
+      "grad_norm": 0.8460028260574683,
+      "learning_rate": 2.9473684210526317e-05,
+      "loss": 0.7811,
+      "step": 21
+    },
+    {
+      "epoch": 0.19193020719738277,
+      "grad_norm": 0.6598083455923888,
+      "learning_rate": 3.087719298245614e-05,
+      "loss": 0.7643,
+      "step": 22
+    },
+    {
+      "epoch": 0.20065430752453653,
+      "grad_norm": 0.7215762522174539,
+      "learning_rate": 3.228070175438597e-05,
+      "loss": 0.7617,
+      "step": 23
+    },
+    {
+      "epoch": 0.2093784078516903,
+      "grad_norm": 0.5554406103693584,
+      "learning_rate": 3.368421052631579e-05,
+      "loss": 0.7576,
+      "step": 24
+    },
+    {
+      "epoch": 0.21810250817884405,
+      "grad_norm": 0.6258543895787715,
+      "learning_rate": 3.508771929824562e-05,
+      "loss": 0.7629,
+      "step": 25
+    },
+    {
+      "epoch": 0.2268266085059978,
+      "grad_norm": 0.7676778822568288,
+      "learning_rate": 3.649122807017544e-05,
+      "loss": 0.7521,
+      "step": 26
+    },
+    {
+      "epoch": 0.23555070883315157,
+      "grad_norm": 1.0032530689824866,
+      "learning_rate": 3.789473684210526e-05,
+      "loss": 0.7384,
+      "step": 27
+    },
+    {
+      "epoch": 0.24427480916030533,
+      "grad_norm": 1.2535316239757839,
+      "learning_rate": 3.929824561403509e-05,
+      "loss": 0.7474,
+      "step": 28
+    },
+    {
+      "epoch": 0.2529989094874591,
+      "grad_norm": 0.9953077887554728,
+      "learning_rate": 4.070175438596492e-05,
+      "loss": 0.7412,
+      "step": 29
+    },
+    {
+      "epoch": 0.2617230098146129,
+      "grad_norm": 1.3600760898966002,
+      "learning_rate": 4.210526315789474e-05,
+      "loss": 0.7257,
+      "step": 30
+    },
+    {
+      "epoch": 0.27044711014176664,
+      "grad_norm": 0.6326945767090465,
+      "learning_rate": 4.350877192982457e-05,
+      "loss": 0.7247,
+      "step": 31
+    },
+    {
+      "epoch": 0.2791712104689204,
+      "grad_norm": 1.0488194881380388,
+      "learning_rate": 4.491228070175439e-05,
+      "loss": 0.7247,
+      "step": 32
+    },
+    {
+      "epoch": 0.28789531079607417,
+      "grad_norm": 1.1854260719924898,
+      "learning_rate": 4.6315789473684214e-05,
+      "loss": 0.7265,
+      "step": 33
+    },
+    {
+      "epoch": 0.2966194111232279,
+      "grad_norm": 1.231800504978088,
+      "learning_rate": 4.771929824561404e-05,
+      "loss": 0.724,
+      "step": 34
+    },
+    {
+      "epoch": 0.3053435114503817,
+      "grad_norm": 1.1471910683800164,
+      "learning_rate": 4.9122807017543864e-05,
+      "loss": 0.7245,
+      "step": 35
+    },
+    {
+      "epoch": 0.31406761177753545,
+      "grad_norm": 1.2650375204215663,
+      "learning_rate": 5.052631578947368e-05,
+      "loss": 0.7173,
+      "step": 36
+    },
+    {
+      "epoch": 0.3227917121046892,
+      "grad_norm": 1.581708517443491,
+      "learning_rate": 5.1929824561403515e-05,
+      "loss": 0.7146,
+      "step": 37
+    },
+    {
+      "epoch": 0.33151581243184297,
+      "grad_norm": 0.6899231907687099,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 0.7039,
+      "step": 38
+    },
+    {
+      "epoch": 0.34023991275899673,
+      "grad_norm": 1.9659650918585876,
+      "learning_rate": 5.4736842105263165e-05,
+      "loss": 0.7129,
+      "step": 39
+    },
+    {
+      "epoch": 0.3489640130861505,
+      "grad_norm": 0.9602545263207618,
+      "learning_rate": 5.6140350877192984e-05,
+      "loss": 0.7229,
+      "step": 40
+    },
+    {
+      "epoch": 0.35768811341330425,
+      "grad_norm": 2.019396688493347,
+      "learning_rate": 5.7543859649122816e-05,
+      "loss": 0.7173,
+      "step": 41
+    },
+    {
+      "epoch": 0.366412213740458,
+      "grad_norm": 1.3667216071947406,
+      "learning_rate": 5.8947368421052634e-05,
+      "loss": 0.7088,
+      "step": 42
+    },
+    {
+      "epoch": 0.3751363140676118,
+      "grad_norm": 1.6739220492126983,
+      "learning_rate": 6.035087719298246e-05,
+      "loss": 0.7126,
+      "step": 43
+    },
+    {
+      "epoch": 0.38386041439476554,
+      "grad_norm": 1.406631076781101,
+      "learning_rate": 6.175438596491228e-05,
+      "loss": 0.7211,
+      "step": 44
+    },
+    {
+      "epoch": 0.3925845147219193,
+      "grad_norm": 1.232815581243198,
+      "learning_rate": 6.315789473684212e-05,
+      "loss": 0.7114,
+      "step": 45
+    },
+    {
+      "epoch": 0.40130861504907306,
+      "grad_norm": 1.3716611381630202,
+      "learning_rate": 6.456140350877194e-05,
+      "loss": 0.7174,
+      "step": 46
+    },
+    {
+      "epoch": 0.4100327153762268,
+      "grad_norm": 1.1415702503019296,
+      "learning_rate": 6.596491228070175e-05,
+      "loss": 0.7005,
+      "step": 47
+    },
+    {
+      "epoch": 0.4187568157033806,
+      "grad_norm": 1.2074860027919303,
+      "learning_rate": 6.736842105263159e-05,
+      "loss": 0.7098,
+      "step": 48
+    },
+    {
+      "epoch": 0.42748091603053434,
+      "grad_norm": 1.6609425790213797,
+      "learning_rate": 6.87719298245614e-05,
+      "loss": 0.7098,
+      "step": 49
+    },
+    {
+      "epoch": 0.4362050163576881,
+      "grad_norm": 0.9769451874458346,
+      "learning_rate": 7.017543859649124e-05,
+      "loss": 0.7079,
+      "step": 50
+    },
+    {
+      "epoch": 0.44492911668484186,
+      "grad_norm": 1.7666355593904801,
+      "learning_rate": 7.157894736842105e-05,
+      "loss": 0.6967,
+      "step": 51
+    },
+    {
+      "epoch": 0.4536532170119956,
+      "grad_norm": 1.014343915032852,
+      "learning_rate": 7.298245614035087e-05,
+      "loss": 0.6961,
+      "step": 52
+    },
+    {
+      "epoch": 0.4623773173391494,
+      "grad_norm": 1.292912081295997,
+      "learning_rate": 7.43859649122807e-05,
+      "loss": 0.6888,
+      "step": 53
+    },
+    {
+      "epoch": 0.47110141766630315,
+      "grad_norm": 1.33549845587438,
+      "learning_rate": 7.578947368421052e-05,
+      "loss": 0.6989,
+      "step": 54
+    },
+    {
+      "epoch": 0.4798255179934569,
+      "grad_norm": 1.6285853466115705,
+      "learning_rate": 7.719298245614036e-05,
+      "loss": 0.7016,
+      "step": 55
+    },
+    {
+      "epoch": 0.48854961832061067,
+      "grad_norm": 0.9433375871219883,
+      "learning_rate": 7.859649122807017e-05,
+      "loss": 0.6824,
+      "step": 56
+    },
+    {
+      "epoch": 0.49727371864776443,
+      "grad_norm": 1.2489894514867836,
+      "learning_rate": 8e-05,
+      "loss": 0.6898,
+      "step": 57
+    },
+    {
+      "epoch": 0.5059978189749182,
+      "grad_norm": 1.60283780700127,
+      "learning_rate": 7.99992499440621e-05,
+      "loss": 0.7019,
+      "step": 58
+    },
+    {
+      "epoch": 0.514721919302072,
+      "grad_norm": 1.349411337094158,
+      "learning_rate": 7.999699980437755e-05,
+      "loss": 0.692,
+      "step": 59
+    },
+    {
+      "epoch": 0.5234460196292258,
+      "grad_norm": 1.4027794489135887,
+      "learning_rate": 7.999324966533291e-05,
+      "loss": 0.6739,
+      "step": 60
+    },
+    {
+      "epoch": 0.5321701199563795,
+      "grad_norm": 0.8999831714962346,
+      "learning_rate": 7.998799966756889e-05,
+      "loss": 0.6925,
+      "step": 61
+    },
+    {
+      "epoch": 0.5408942202835333,
+      "grad_norm": 1.2349556182219792,
+      "learning_rate": 7.998125000797506e-05,
+      "loss": 0.6811,
+      "step": 62
+    },
+    {
+      "epoch": 0.549618320610687,
+      "grad_norm": 0.9953482513645127,
+      "learning_rate": 7.997300093968255e-05,
+      "loss": 0.6919,
+      "step": 63
+    },
+    {
+      "epoch": 0.5583424209378408,
+      "grad_norm": 1.7631278928848257,
+      "learning_rate": 7.99632527720545e-05,
+      "loss": 0.7076,
+      "step": 64
+    },
+    {
+      "epoch": 0.5670665212649946,
+      "grad_norm": 0.9802682486455654,
+      "learning_rate": 7.995200587067445e-05,
+      "loss": 0.6836,
+      "step": 65
+    },
+    {
+      "epoch": 0.5757906215921483,
+      "grad_norm": 1.4018686830735987,
+      "learning_rate": 7.993926065733265e-05,
+      "loss": 0.6998,
+      "step": 66
+    },
+    {
+      "epoch": 0.5845147219193021,
+      "grad_norm": 1.0081781512305683,
+      "learning_rate": 7.992501761001027e-05,
+      "loss": 0.6878,
+      "step": 67
+    },
+    {
+      "epoch": 0.5932388222464559,
+      "grad_norm": 1.2813118135682324,
+      "learning_rate": 7.99092772628614e-05,
+      "loss": 0.6948,
+      "step": 68
+    },
+    {
+      "epoch": 0.6019629225736096,
+      "grad_norm": 0.9281112795468938,
+      "learning_rate": 7.98920402061931e-05,
+      "loss": 0.6969,
+      "step": 69
+    },
+    {
+      "epoch": 0.6106870229007634,
+      "grad_norm": 1.2375443626494387,
+      "learning_rate": 7.987330708644319e-05,
+      "loss": 0.6793,
+      "step": 70
+    },
+    {
+      "epoch": 0.6194111232279171,
+      "grad_norm": 1.0863449033166035,
+      "learning_rate": 7.985307860615607e-05,
+      "loss": 0.6848,
+      "step": 71
+    },
+    {
+      "epoch": 0.6281352235550709,
+      "grad_norm": 0.942362107324505,
+      "learning_rate": 7.98313555239563e-05,
+      "loss": 0.6862,
+      "step": 72
+    },
+    {
+      "epoch": 0.6368593238822247,
+      "grad_norm": 1.3027161650276327,
+      "learning_rate": 7.980813865452026e-05,
+      "loss": 0.6735,
+      "step": 73
+    },
+    {
+      "epoch": 0.6455834242093784,
+      "grad_norm": 0.7977961043410647,
+      "learning_rate": 7.978342886854546e-05,
+      "loss": 0.6787,
+      "step": 74
+    },
+    {
+      "epoch": 0.6543075245365322,
+      "grad_norm": 1.3693856212692685,
+      "learning_rate": 7.975722709271799e-05,
+      "loss": 0.6751,
+      "step": 75
+    },
+    {
+      "epoch": 0.6630316248636859,
+      "grad_norm": 0.7738139399748106,
+      "learning_rate": 7.972953430967773e-05,
+      "loss": 0.6726,
+      "step": 76
+    },
+    {
+      "epoch": 0.6717557251908397,
+      "grad_norm": 0.6453937973753266,
+      "learning_rate": 7.97003515579815e-05,
+      "loss": 0.6756,
+      "step": 77
+    },
+    {
+      "epoch": 0.6804798255179935,
+      "grad_norm": 0.7085908664971486,
+      "learning_rate": 7.96696799320641e-05,
+      "loss": 0.6774,
+      "step": 78
+    },
+    {
+      "epoch": 0.6892039258451472,
+      "grad_norm": 0.7420105586344222,
+      "learning_rate": 7.96375205821973e-05,
+      "loss": 0.6647,
+      "step": 79
+    },
+    {
+      "epoch": 0.697928026172301,
+      "grad_norm": 1.182903610639294,
+      "learning_rate": 7.960387471444666e-05,
+      "loss": 0.6758,
+      "step": 80
+    },
+    {
+      "epoch": 0.7066521264994547,
+      "grad_norm": 1.305515943296027,
+      "learning_rate": 7.956874359062632e-05,
+      "loss": 0.6808,
+      "step": 81
+    },
+    {
+      "epoch": 0.7153762268266085,
+      "grad_norm": 0.9451990102208602,
+      "learning_rate": 7.95321285282517e-05,
+      "loss": 0.6697,
+      "step": 82
+    },
+    {
+      "epoch": 0.7241003271537623,
+      "grad_norm": 0.9432391308082473,
+      "learning_rate": 7.949403090049002e-05,
+      "loss": 0.6723,
+      "step": 83
+    },
+    {
+      "epoch": 0.732824427480916,
+      "grad_norm": 1.1371193647756934,
+      "learning_rate": 7.94544521361089e-05,
+      "loss": 0.6773,
+      "step": 84
+    },
+    {
+      "epoch": 0.7415485278080698,
+      "grad_norm": 1.1858932832021734,
+      "learning_rate": 7.941339371942269e-05,
+      "loss": 0.6657,
+      "step": 85
+    },
+    {
+      "epoch": 0.7502726281352236,
+      "grad_norm": 1.165847606968587,
+      "learning_rate": 7.937085719023685e-05,
+      "loss": 0.6743,
+      "step": 86
+    },
+    {
+      "epoch": 0.7589967284623773,
+      "grad_norm": 0.5770889497138026,
+      "learning_rate": 7.932684414379021e-05,
+      "loss": 0.668,
+      "step": 87
+    },
+    {
+      "epoch": 0.7677208287895311,
+      "grad_norm": 0.925349798073699,
+      "learning_rate": 7.928135623069509e-05,
+      "loss": 0.6693,
+      "step": 88
+    },
+    {
+      "epoch": 0.7764449291166848,
+      "grad_norm": 1.2265564072341149,
+      "learning_rate": 7.923439515687546e-05,
+      "loss": 0.6721,
+      "step": 89
+    },
+    {
+      "epoch": 0.7851690294438386,
+      "grad_norm": 0.76774338811776,
+      "learning_rate": 7.918596268350296e-05,
+      "loss": 0.664,
+      "step": 90
+    },
+    {
+      "epoch": 0.7938931297709924,
+      "grad_norm": 0.811830002272808,
+      "learning_rate": 7.913606062693077e-05,
+      "loss": 0.6674,
+      "step": 91
+    },
+    {
+      "epoch": 0.8026172300981461,
+      "grad_norm": 0.7896885933612465,
+      "learning_rate": 7.90846908586256e-05,
+      "loss": 0.6603,
+      "step": 92
+    },
+    {
+      "epoch": 0.8113413304252999,
+      "grad_norm": 0.6860009351157326,
+      "learning_rate": 7.903185530509743e-05,
+      "loss": 0.6541,
+      "step": 93
+    },
+    {
+      "epoch": 0.8200654307524536,
+      "grad_norm": 0.7887179980130798,
+      "learning_rate": 7.89775559478273e-05,
+      "loss": 0.6719,
+      "step": 94
+    },
+    {
+      "epoch": 0.8287895310796074,
+      "grad_norm": 0.6863156036434349,
+      "learning_rate": 7.892179482319297e-05,
+      "loss": 0.6712,
+      "step": 95
+    },
+    {
+      "epoch": 0.8375136314067612,
+      "grad_norm": 0.612318445743932,
+      "learning_rate": 7.886457402239256e-05,
+      "loss": 0.6588,
+      "step": 96
+    },
+    {
+      "epoch": 0.8462377317339149,
+      "grad_norm": 0.5266659568320967,
+      "learning_rate": 7.880589569136616e-05,
+      "loss": 0.6477,
+      "step": 97
+    },
+    {
+      "epoch": 0.8549618320610687,
+      "grad_norm": 0.6815563256417184,
+      "learning_rate": 7.874576203071531e-05,
+      "loss": 0.6647,
+      "step": 98
+    },
+    {
+      "epoch": 0.8636859323882224,
+      "grad_norm": 0.7021771362356389,
+      "learning_rate": 7.868417529562043e-05,
+      "loss": 0.6519,
+      "step": 99
+    },
+    {
+      "epoch": 0.8724100327153762,
+      "grad_norm": 0.9537663707386643,
+      "learning_rate": 7.862113779575638e-05,
+      "loss": 0.6655,
+      "step": 100
+    },
+    {
+      "epoch": 0.88113413304253,
+      "grad_norm": 1.7262458672727332,
+      "learning_rate": 7.85566518952057e-05,
+      "loss": 0.6617,
+      "step": 101
+    },
+    {
+      "epoch": 0.8898582333696837,
+      "grad_norm": 0.5924197078630852,
+      "learning_rate": 7.849072001237001e-05,
+      "loss": 0.6586,
+      "step": 102
+    },
+    {
+      "epoch": 0.8985823336968375,
+      "grad_norm": 1.784429399577841,
+      "learning_rate": 7.842334461987936e-05,
+      "loss": 0.6718,
+      "step": 103
+    },
+    {
+      "epoch": 0.9073064340239912,
+      "grad_norm": 0.8548005705828143,
+      "learning_rate": 7.835452824449935e-05,
+      "loss": 0.6589,
+      "step": 104
+    },
+    {
+      "epoch": 0.916030534351145,
+      "grad_norm": 1.5914313223828267,
+      "learning_rate": 7.828427346703657e-05,
+      "loss": 0.6639,
+      "step": 105
+    },
+    {
+      "epoch": 0.9247546346782988,
+      "grad_norm": 1.1912921777513241,
+      "learning_rate": 7.821258292224166e-05,
+      "loss": 0.669,
+      "step": 106
+    },
+    {
+      "epoch": 0.9334787350054525,
+      "grad_norm": 1.1951669530416693,
+      "learning_rate": 7.813945929871056e-05,
+      "loss": 0.6497,
+      "step": 107
+    },
+    {
+      "epoch": 0.9422028353326063,
+      "grad_norm": 1.037062568022807,
+      "learning_rate": 7.806490533878368e-05,
+      "loss": 0.6632,
+      "step": 108
+    },
+    {
+      "epoch": 0.95092693565976,
+      "grad_norm": 1.2483557427639183,
+      "learning_rate": 7.798892383844303e-05,
+      "loss": 0.658,
+      "step": 109
+    },
+    {
+      "epoch": 0.9596510359869138,
+      "grad_norm": 0.8133626652288988,
+      "learning_rate": 7.791151764720737e-05,
+      "loss": 0.6571,
+      "step": 110
+    },
+    {
+      "epoch": 0.9683751363140676,
+      "grad_norm": 1.0059615445006975,
+      "learning_rate": 7.783268966802539e-05,
+      "loss": 0.6595,
+      "step": 111
+    },
+    {
+      "epoch": 0.9770992366412213,
+      "grad_norm": 0.7304172503042858,
+      "learning_rate": 7.775244285716679e-05,
+      "loss": 0.6608,
+      "step": 112
+    },
+    {
+      "epoch": 0.9858233369683751,
+      "grad_norm": 0.7284490288054312,
+      "learning_rate": 7.767078022411139e-05,
+      "loss": 0.6492,
+      "step": 113
+    },
+    {
+      "epoch": 0.9945474372955289,
+      "grad_norm": 0.5901533647047473,
+      "learning_rate": 7.758770483143634e-05,
+      "loss": 0.6642,
+      "step": 114
+    },
+    {
+      "epoch": 1.0038167938931297,
+      "grad_norm": 0.8414397249541458,
+      "learning_rate": 7.750321979470123e-05,
+      "loss": 0.9152,
+      "step": 115
+    },
+    {
+      "epoch": 1.0125408942202836,
+      "grad_norm": 1.0739193248054102,
+      "learning_rate": 7.741732828233124e-05,
+      "loss": 0.626,
+      "step": 116
+    },
+    {
+      "epoch": 1.0212649945474372,
+      "grad_norm": 1.1988463490729344,
+      "learning_rate": 7.733003351549829e-05,
+      "loss": 0.6393,
+      "step": 117
+    },
+    {
+      "epoch": 1.0299890948745911,
+      "grad_norm": 0.9606655160302263,
+      "learning_rate": 7.724133876800031e-05,
+      "loss": 0.625,
+      "step": 118
+    },
+    {
+      "epoch": 1.0387131952017448,
+      "grad_norm": 1.2372012961246242,
+      "learning_rate": 7.715124736613839e-05,
+      "loss": 0.6278,
+      "step": 119
+    },
+    {
+      "epoch": 1.0474372955288986,
+      "grad_norm": 0.86001815183547,
+      "learning_rate": 7.705976268859207e-05,
+      "loss": 0.6187,
+      "step": 120
+    },
+    {
+      "epoch": 1.0561613958560523,
+      "grad_norm": 1.1413833226380308,
+      "learning_rate": 7.696688816629266e-05,
+      "loss": 0.63,
+      "step": 121
+    },
+    {
+      "epoch": 1.0648854961832062,
+      "grad_norm": 1.2327702775009692,
+      "learning_rate": 7.687262728229447e-05,
+      "loss": 0.6205,
+      "step": 122
+    },
+    {
+      "epoch": 1.0736095965103598,
+      "grad_norm": 0.7767212251506833,
+      "learning_rate": 7.677698357164431e-05,
+      "loss": 0.6177,
+      "step": 123
+    },
+    {
+      "epoch": 1.0823336968375137,
+      "grad_norm": 1.7079661807939999,
+      "learning_rate": 7.667996062124884e-05,
+      "loss": 0.6242,
+      "step": 124
+    },
+    {
+      "epoch": 1.0910577971646673,
+      "grad_norm": 1.0068734700150146,
+      "learning_rate": 7.658156206974005e-05,
+      "loss": 0.6183,
+      "step": 125
+    },
+    {
+      "epoch": 1.0997818974918212,
+      "grad_norm": 1.8753922872596245,
+      "learning_rate": 7.648179160733883e-05,
+      "loss": 0.6337,
+      "step": 126
+    },
+    {
+      "epoch": 1.1085059978189749,
+      "grad_norm": 1.8578444866986128,
+      "learning_rate": 7.638065297571657e-05,
+      "loss": 0.6171,
+      "step": 127
+    },
+    {
+      "epoch": 1.1172300981461287,
+      "grad_norm": 0.6478551153484415,
+      "learning_rate": 7.627814996785484e-05,
+      "loss": 0.6158,
+      "step": 128
+    },
+    {
+      "epoch": 1.1259541984732824,
+      "grad_norm": 1.4820468569461809,
+      "learning_rate": 7.61742864279031e-05,
+      "loss": 0.6197,
+      "step": 129
+    },
+    {
+      "epoch": 1.1346782988004362,
+      "grad_norm": 0.7243487925020168,
+      "learning_rate": 7.606906625103464e-05,
+      "loss": 0.625,
+      "step": 130
+    },
+    {
+      "epoch": 1.14340239912759,
+      "grad_norm": 0.9539104928282168,
+      "learning_rate": 7.596249338330034e-05,
+      "loss": 0.6194,
+      "step": 131
+    },
+    {
+      "epoch": 1.1521264994547438,
+      "grad_norm": 0.8136844990028549,
+      "learning_rate": 7.585457182148081e-05,
+      "loss": 0.6172,
+      "step": 132
+    },
+    {
+      "epoch": 1.1608505997818974,
+      "grad_norm": 0.7125736106844306,
+      "learning_rate": 7.57453056129365e-05,
+      "loss": 0.6151,
+      "step": 133
+    },
+    {
+      "epoch": 1.1695747001090513,
+      "grad_norm": 0.5804990189493681,
+      "learning_rate": 7.56346988554558e-05,
+      "loss": 0.6256,
+      "step": 134
+    },
+    {
+      "epoch": 1.178298800436205,
+      "grad_norm": 0.6362204084540434,
+      "learning_rate": 7.552275569710152e-05,
+      "loss": 0.6184,
+      "step": 135
+    },
+    {
+      "epoch": 1.1870229007633588,
+      "grad_norm": 0.43420077191532125,
+      "learning_rate": 7.540948033605513e-05,
+      "loss": 0.6105,
+      "step": 136
+    },
+    {
+      "epoch": 1.1957470010905125,
+      "grad_norm": 0.46232316550703767,
+      "learning_rate": 7.529487702045953e-05,
+      "loss": 0.6056,
+      "step": 137
+    },
+    {
+      "epoch": 1.2044711014176663,
+      "grad_norm": 0.4316498267915253,
+      "learning_rate": 7.517895004825956e-05,
+      "loss": 0.6011,
+      "step": 138
+    },
+    {
+      "epoch": 1.21319520174482,
+      "grad_norm": 0.38696383075278273,
+      "learning_rate": 7.506170376704095e-05,
+      "loss": 0.6141,
+      "step": 139
+    },
+    {
+      "epoch": 1.2219193020719739,
+      "grad_norm": 0.3858939834475368,
+      "learning_rate": 7.494314257386715e-05,
+      "loss": 0.6065,
+      "step": 140
+    },
+    {
+      "epoch": 1.2306434023991275,
+      "grad_norm": 0.3738279653817542,
+      "learning_rate": 7.48232709151145e-05,
+      "loss": 0.6104,
+      "step": 141
+    },
+    {
+      "epoch": 1.2393675027262814,
+      "grad_norm": 0.34434724315715803,
+      "learning_rate": 7.470209328630548e-05,
+      "loss": 0.605,
+      "step": 142
+    },
+    {
+      "epoch": 1.248091603053435,
+      "grad_norm": 0.3291602472493041,
+      "learning_rate": 7.457961423194011e-05,
+      "loss": 0.6018,
+      "step": 143
+    },
+    {
+      "epoch": 1.256815703380589,
+      "grad_norm": 0.27837095142325796,
+      "learning_rate": 7.445583834532546e-05,
+      "loss": 0.5993,
+      "step": 144
+    },
+    {
+      "epoch": 1.2655398037077425,
+      "grad_norm": 0.31225187743500077,
+      "learning_rate": 7.433077026840346e-05,
+      "loss": 0.5984,
+      "step": 145
+    },
+    {
+      "epoch": 1.2742639040348964,
+      "grad_norm": 0.27063932853508044,
+      "learning_rate": 7.420441469157684e-05,
+      "loss": 0.6048,
+      "step": 146
+    },
+    {
+      "epoch": 1.2829880043620503,
+      "grad_norm": 0.28328810560800655,
+      "learning_rate": 7.407677635353308e-05,
+      "loss": 0.6061,
+      "step": 147
+    },
+    {
+      "epoch": 1.291712104689204,
+      "grad_norm": 0.3335236260280515,
+      "learning_rate": 7.39478600410669e-05,
+      "loss": 0.6055,
+      "step": 148
+    },
+    {
+      "epoch": 1.3004362050163576,
+      "grad_norm": 0.34277641090811606,
+      "learning_rate": 7.381767058890056e-05,
+      "loss": 0.5956,
+      "step": 149
+    },
+    {
+      "epoch": 1.3091603053435115,
+      "grad_norm": 0.3944860323036086,
+      "learning_rate": 7.368621287950264e-05,
+      "loss": 0.5971,
+      "step": 150
+    },
+    {
+      "epoch": 1.3178844056706653,
+      "grad_norm": 0.390412169412763,
+      "learning_rate": 7.355349184290491e-05,
+      "loss": 0.5991,
+      "step": 151
+    },
+    {
+      "epoch": 1.326608505997819,
+      "grad_norm": 0.44167362977698205,
+      "learning_rate": 7.341951245651747e-05,
+      "loss": 0.5905,
+      "step": 152
+    },
+    {
+      "epoch": 1.3353326063249726,
+      "grad_norm": 0.4873733423530836,
+      "learning_rate": 7.328427974494201e-05,
+      "loss": 0.6061,
+      "step": 153
+    },
+    {
+      "epoch": 1.3440567066521265,
+      "grad_norm": 0.529005701508295,
+      "learning_rate": 7.314779877978346e-05,
+      "loss": 0.6038,
+      "step": 154
+    },
+    {
+      "epoch": 1.3527808069792804,
+      "grad_norm": 0.5776151078666283,
+      "learning_rate": 7.301007467945974e-05,
+      "loss": 0.6063,
+      "step": 155
+    },
+    {
+      "epoch": 1.361504907306434,
+      "grad_norm": 0.6712943884189805,
+      "learning_rate": 7.28711126090098e-05,
+      "loss": 0.609,
+      "step": 156
+    },
+    {
+      "epoch": 1.3702290076335877,
+      "grad_norm": 0.6974077084236574,
+      "learning_rate": 7.273091777989997e-05,
+      "loss": 0.617,
+      "step": 157
+    },
+    {
+      "epoch": 1.3789531079607416,
+      "grad_norm": 0.622800075821039,
+      "learning_rate": 7.258949544982843e-05,
+      "loss": 0.6012,
+      "step": 158
+    },
+    {
+      "epoch": 1.3876772082878954,
+      "grad_norm": 0.6652165160394861,
+      "learning_rate": 7.24468509225281e-05,
+      "loss": 0.6213,
+      "step": 159
+    },
+    {
+      "epoch": 1.396401308615049,
+      "grad_norm": 0.7796997637986155,
+      "learning_rate": 7.230298954756772e-05,
+      "loss": 0.6116,
+      "step": 160
+    },
+    {
+      "epoch": 1.4051254089422027,
+      "grad_norm": 0.8050974173338423,
+      "learning_rate": 7.215791672015121e-05,
+      "loss": 0.6102,
+      "step": 161
+    },
+    {
+      "epoch": 1.4138495092693566,
+      "grad_norm": 0.6245666904233175,
+      "learning_rate": 7.201163788091536e-05,
+      "loss": 0.6051,
+      "step": 162
+    },
+    {
+      "epoch": 1.4225736095965105,
+      "grad_norm": 0.2791815826660419,
+      "learning_rate": 7.186415851572579e-05,
+      "loss": 0.5978,
+      "step": 163
+    },
+    {
+      "epoch": 1.4312977099236641,
+      "grad_norm": 0.5658363527146323,
+      "learning_rate": 7.171548415547114e-05,
+      "loss": 0.6064,
+      "step": 164
+    },
+    {
+      "epoch": 1.4400218102508178,
+      "grad_norm": 0.6010202995307374,
+      "learning_rate": 7.156562037585576e-05,
+      "loss": 0.6117,
+      "step": 165
+    },
+    {
+      "epoch": 1.4487459105779716,
+      "grad_norm": 0.45545843396243296,
+      "learning_rate": 7.141457279719053e-05,
+      "loss": 0.6079,
+      "step": 166
+    },
+    {
+      "epoch": 1.4574700109051255,
+      "grad_norm": 0.37879238626475686,
+      "learning_rate": 7.126234708418214e-05,
+      "loss": 0.5965,
+      "step": 167
+    },
+    {
+      "epoch": 1.4661941112322792,
+      "grad_norm": 0.4580319723211172,
+      "learning_rate": 7.110894894572056e-05,
+      "loss": 0.6106,
+      "step": 168
+    },
+    {
+      "epoch": 1.4749182115594328,
+      "grad_norm": 0.43310826738894753,
+      "learning_rate": 7.095438413466503e-05,
+      "loss": 0.6152,
+      "step": 169
+    },
+    {
+      "epoch": 1.4836423118865867,
+      "grad_norm": 0.42673749198290256,
+      "learning_rate": 7.079865844762829e-05,
+      "loss": 0.6058,
+      "step": 170
+    },
+    {
+      "epoch": 1.4923664122137406,
+      "grad_norm": 0.5356882769390818,
+      "learning_rate": 7.064177772475912e-05,
+      "loss": 0.6011,
+      "step": 171
+    },
+    {
+      "epoch": 1.5010905125408942,
+      "grad_norm": 0.48453744479417066,
+      "learning_rate": 7.048374784952343e-05,
+      "loss": 0.6014,
+      "step": 172
+    },
+    {
+      "epoch": 1.5098146128680479,
+      "grad_norm": 0.40573397034428954,
+      "learning_rate": 7.03245747484835e-05,
+      "loss": 0.6021,
+      "step": 173
+    },
+    {
+      "epoch": 1.5185387131952017,
+      "grad_norm": 0.5052592973873418,
+      "learning_rate": 7.016426439107586e-05,
+      "loss": 0.5976,
+      "step": 174
+    },
+    {
+      "epoch": 1.5272628135223556,
+      "grad_norm": 0.5640845358828827,
+      "learning_rate": 7.000282278938724e-05,
+      "loss": 0.6032,
+      "step": 175
+    },
+    {
+      "epoch": 1.5359869138495092,
+      "grad_norm": 0.5716127749488464,
+      "learning_rate": 6.984025599792926e-05,
+      "loss": 0.6069,
+      "step": 176
+    },
+    {
+      "epoch": 1.544711014176663,
+      "grad_norm": 0.6147245538888289,
+      "learning_rate": 6.967657011341126e-05,
+      "loss": 0.6017,
+      "step": 177
+    },
+    {
+      "epoch": 1.5534351145038168,
+      "grad_norm": 0.5972264181543786,
+      "learning_rate": 6.951177127451177e-05,
+      "loss": 0.6039,
+      "step": 178
+    },
+    {
+      "epoch": 1.5621592148309706,
+      "grad_norm": 0.4511378630106592,
+      "learning_rate": 6.934586566164811e-05,
+      "loss": 0.609,
+      "step": 179
+    },
+    {
+      "epoch": 1.5708833151581243,
+      "grad_norm": 0.3324255798317227,
+      "learning_rate": 6.917885949674483e-05,
+      "loss": 0.6046,
+      "step": 180
+    },
+    {
+      "epoch": 1.579607415485278,
+      "grad_norm": 0.3318537793653575,
+      "learning_rate": 6.901075904300021e-05,
+      "loss": 0.5984,
+      "step": 181
+    },
+    {
+      "epoch": 1.5883315158124318,
+      "grad_norm": 0.4025671963472955,
+      "learning_rate": 6.88415706046514e-05,
+      "loss": 0.5991,
+      "step": 182
+    },
+    {
+      "epoch": 1.5970556161395857,
+      "grad_norm": 0.4039375325984613,
+      "learning_rate": 6.867130052673806e-05,
+      "loss": 0.6132,
+      "step": 183
+    },
+    {
+      "epoch": 1.6057797164667393,
+      "grad_norm": 0.42633324449837107,
+      "learning_rate": 6.849995519486434e-05,
+      "loss": 0.6112,
+      "step": 184
+    },
+    {
+      "epoch": 1.614503816793893,
+      "grad_norm": 0.4031771610559183,
+      "learning_rate": 6.832754103495939e-05,
+      "loss": 0.5951,
+      "step": 185
+    },
+    {
+      "epoch": 1.6232279171210469,
+      "grad_norm": 0.3101345676851429,
+      "learning_rate": 6.815406451303647e-05,
+      "loss": 0.5977,
+      "step": 186
+    },
+    {
+      "epoch": 1.6319520174482007,
+      "grad_norm": 0.28180350855331504,
+      "learning_rate": 6.797953213495033e-05,
+      "loss": 0.6081,
+      "step": 187
+    },
+    {
+      "epoch": 1.6406761177753544,
+      "grad_norm": 0.3450910788047459,
+      "learning_rate": 6.780395044615329e-05,
+      "loss": 0.606,
+      "step": 188
+    },
+    {
+      "epoch": 1.649400218102508,
+      "grad_norm": 0.3941732240418325,
+      "learning_rate": 6.762732603144978e-05,
+      "loss": 0.6079,
+      "step": 189
+    },
+    {
+      "epoch": 1.658124318429662,
+      "grad_norm": 0.3241938707733741,
+      "learning_rate": 6.744966551474936e-05,
+      "loss": 0.604,
+      "step": 190
+    },
+    {
+      "epoch": 1.6668484187568158,
+      "grad_norm": 0.27804934781744717,
+      "learning_rate": 6.727097555881826e-05,
+      "loss": 0.595,
+      "step": 191
+    },
+    {
+      "epoch": 1.6755725190839694,
+      "grad_norm": 0.25336159667743025,
+      "learning_rate": 6.709126286502965e-05,
+      "loss": 0.6025,
+      "step": 192
+    },
+    {
+      "epoch": 1.684296619411123,
+      "grad_norm": 0.29341065316587794,
+      "learning_rate": 6.691053417311216e-05,
+      "loss": 0.6023,
+      "step": 193
+    },
+    {
+      "epoch": 1.693020719738277,
+      "grad_norm": 0.2878985409168238,
+      "learning_rate": 6.672879626089723e-05,
+      "loss": 0.5949,
+      "step": 194
+    },
+    {
+      "epoch": 1.7017448200654308,
+      "grad_norm": 0.26927072720134837,
+      "learning_rate": 6.654605594406486e-05,
+      "loss": 0.5979,
+      "step": 195
+    },
+    {
+      "epoch": 1.7104689203925845,
+      "grad_norm": 0.3155329601805801,
+      "learning_rate": 6.636232007588805e-05,
+      "loss": 0.587,
+      "step": 196
+    },
+    {
+      "epoch": 1.7191930207197381,
+      "grad_norm": 0.34847815991034525,
+      "learning_rate": 6.617759554697573e-05,
+      "loss": 0.5978,
+      "step": 197
+    },
+    {
+      "epoch": 1.727917121046892,
+      "grad_norm": 0.28896074500408486,
+      "learning_rate": 6.59918892850144e-05,
+      "loss": 0.6069,
+      "step": 198
+    },
+    {
+      "epoch": 1.7366412213740459,
+      "grad_norm": 0.2981626524269195,
+      "learning_rate": 6.580520825450827e-05,
+      "loss": 0.5956,
+      "step": 199
+    },
+    {
+      "epoch": 1.7453653217011995,
+      "grad_norm": 0.31680273274222387,
+      "learning_rate": 6.561755945651813e-05,
+      "loss": 0.5907,
+      "step": 200
+    },
+    {
+      "epoch": 1.7540894220283532,
+      "grad_norm": 0.2788023167998957,
+      "learning_rate": 6.542894992839873e-05,
+      "loss": 0.5988,
+      "step": 201
+    },
+    {
+      "epoch": 1.762813522355507,
+      "grad_norm": 0.26041179212976306,
+      "learning_rate": 6.52393867435349e-05,
+      "loss": 0.5879,
+      "step": 202
+    },
+    {
+      "epoch": 1.771537622682661,
+      "grad_norm": 0.34950012098507055,
+      "learning_rate": 6.504887701107626e-05,
+      "loss": 0.5933,
+      "step": 203
+    },
+    {
+      "epoch": 1.7802617230098146,
+      "grad_norm": 0.5298511066656313,
+      "learning_rate": 6.48574278756706e-05,
+      "loss": 0.59,
+      "step": 204
+    },
+    {
+      "epoch": 1.7889858233369684,
+      "grad_norm": 0.6366920336388178,
+      "learning_rate": 6.466504651719598e-05,
+      "loss": 0.5967,
+      "step": 205
+    },
+    {
+      "epoch": 1.7977099236641223,
+      "grad_norm": 0.6299010554780248,
+      "learning_rate": 6.447174015049139e-05,
+      "loss": 0.5974,
+      "step": 206
+    },
+    {
+      "epoch": 1.806434023991276,
+      "grad_norm": 0.562858852871968,
+      "learning_rate": 6.427751602508628e-05,
+      "loss": 0.6057,
+      "step": 207
+    },
+    {
+      "epoch": 1.8151581243184296,
+      "grad_norm": 0.4004595325182879,
+      "learning_rate": 6.408238142492855e-05,
+      "loss": 0.59,
+      "step": 208
+    },
+    {
+      "epoch": 1.8238822246455835,
+      "grad_norm": 0.21584901270572757,
+      "learning_rate": 6.388634366811146e-05,
+      "loss": 0.5943,
+      "step": 209
+    },
+    {
+      "epoch": 1.8326063249727373,
+      "grad_norm": 0.37265049074019335,
+      "learning_rate": 6.368941010659921e-05,
+      "loss": 0.6096,
+      "step": 210
+    },
+    {
+      "epoch": 1.841330425299891,
+      "grad_norm": 0.5267910816953789,
+      "learning_rate": 6.349158812595116e-05,
+      "loss": 0.6053,
+      "step": 211
+    },
+    {
+      "epoch": 1.8500545256270446,
+      "grad_norm": 0.47801721985552514,
+      "learning_rate": 6.329288514504487e-05,
+      "loss": 0.5936,
+      "step": 212
+    },
+    {
+      "epoch": 1.8587786259541985,
+      "grad_norm": 0.27531874942679646,
+      "learning_rate": 6.309330861579786e-05,
+      "loss": 0.5997,
+      "step": 213
+    },
+    {
+      "epoch": 1.8675027262813524,
+      "grad_norm": 0.3657379875213062,
+      "learning_rate": 6.28928660228882e-05,
+      "loss": 0.5963,
+      "step": 214
+    },
+    {
+      "epoch": 1.876226826608506,
+      "grad_norm": 0.3633999933893769,
+      "learning_rate": 6.269156488347372e-05,
+      "loss": 0.5899,
+      "step": 215
+    },
+    {
+      "epoch": 1.8849509269356597,
+      "grad_norm": 0.2855859194806056,
+      "learning_rate": 6.248941274691017e-05,
+      "loss": 0.5963,
+      "step": 216
+    },
+    {
+      "epoch": 1.8936750272628136,
+      "grad_norm": 0.26871535087996673,
+      "learning_rate": 6.228641719446808e-05,
+      "loss": 0.5832,
+      "step": 217
+    },
+    {
+      "epoch": 1.9023991275899674,
+      "grad_norm": 0.26413333351828056,
+      "learning_rate": 6.208258583904841e-05,
+      "loss": 0.5913,
+      "step": 218
+    },
+    {
+      "epoch": 1.911123227917121,
+      "grad_norm": 0.27861019506862306,
+      "learning_rate": 6.18779263248971e-05,
+      "loss": 0.6088,
+      "step": 219
+    },
+    {
+      "epoch": 1.9198473282442747,
+      "grad_norm": 0.26490994643954596,
+      "learning_rate": 6.16724463273183e-05,
+      "loss": 0.6072,
+      "step": 220
+    },
+    {
+      "epoch": 1.9285714285714286,
+      "grad_norm": 0.2464264085957813,
+      "learning_rate": 6.146615355238668e-05,
+      "loss": 0.5998,
+      "step": 221
+    },
+    {
+      "epoch": 1.9372955288985825,
+      "grad_norm": 0.2060880785957384,
+      "learning_rate": 6.125905573665824e-05,
+      "loss": 0.5998,
+      "step": 222
+    },
+    {
+      "epoch": 1.9460196292257361,
+      "grad_norm": 0.26971063557098873,
+      "learning_rate": 6.105116064688033e-05,
+      "loss": 0.6048,
+      "step": 223
+    },
+    {
+      "epoch": 1.9547437295528898,
+      "grad_norm": 0.32136759035233387,
+      "learning_rate": 6.0842476079700264e-05,
+      "loss": 0.6019,
+      "step": 224
+    },
+    {
+      "epoch": 1.9634678298800436,
+      "grad_norm": 0.23286070304886092,
+      "learning_rate": 6.063300986137297e-05,
+      "loss": 0.5978,
+      "step": 225
+    },
+    {
+      "epoch": 1.9721919302071975,
+      "grad_norm": 0.18824642933626035,
+      "learning_rate": 6.04227698474675e-05,
+      "loss": 0.5853,
+      "step": 226
+    },
+    {
+      "epoch": 1.9809160305343512,
+      "grad_norm": 0.21674772391599217,
+      "learning_rate": 6.02117639225724e-05,
+      "loss": 0.6078,
+      "step": 227
+    },
+    {
+      "epoch": 1.9896401308615048,
+      "grad_norm": 0.2934361150620414,
+      "learning_rate": 6.000000000000001e-05,
+      "loss": 0.5984,
+      "step": 228
+    },
+    {
+      "epoch": 1.9983642311886587,
+      "grad_norm": 0.44422300505199064,
+      "learning_rate": 5.9787486021489705e-05,
+      "loss": 0.8056,
+      "step": 229
+    },
+    {
+      "epoch": 2.0076335877862594,
+      "grad_norm": 0.5875332032837016,
+      "learning_rate": 5.957422995691007e-05,
+      "loss": 0.6104,
+      "step": 230
+    },
+    {
+      "epoch": 2.016357688113413,
+      "grad_norm": 0.7504197536874134,
+      "learning_rate": 5.936023980395997e-05,
+      "loss": 0.5445,
+      "step": 231
+    },
+    {
+      "epoch": 2.025081788440567,
+      "grad_norm": 1.1182174595326815,
+      "learning_rate": 5.914552358786864e-05,
+      "loss": 0.5436,
+      "step": 232
+    },
+    {
+      "epoch": 2.033805888767721,
+      "grad_norm": 0.9253408273401474,
+      "learning_rate": 5.893008936109474e-05,
+      "loss": 0.5517,
+      "step": 233
+    },
+    {
+      "epoch": 2.0425299890948745,
+      "grad_norm": 0.6241397091570694,
+      "learning_rate": 5.871394520302432e-05,
+      "loss": 0.5436,
+      "step": 234
+    },
+    {
+      "epoch": 2.051254089422028,
+      "grad_norm": 0.6714967761660577,
+      "learning_rate": 5.8497099219667834e-05,
+      "loss": 0.5513,
+      "step": 235
+    },
+    {
+      "epoch": 2.0599781897491822,
+      "grad_norm": 0.6546289177103435,
+      "learning_rate": 5.827955954335616e-05,
+      "loss": 0.5365,
+      "step": 236
+    },
+    {
+      "epoch": 2.068702290076336,
+      "grad_norm": 0.6213690775351733,
+      "learning_rate": 5.806133433243558e-05,
+      "loss": 0.546,
+      "step": 237
+    },
+    {
+      "epoch": 2.0774263904034895,
+      "grad_norm": 0.6920575535177348,
+      "learning_rate": 5.784243177096187e-05,
+      "loss": 0.5433,
+      "step": 238
+    },
+    {
+      "epoch": 2.0861504907306436,
+      "grad_norm": 0.4645752707045055,
+      "learning_rate": 5.7622860068393334e-05,
+      "loss": 0.5289,
+      "step": 239
+    },
+    {
+      "epoch": 2.0948745910577973,
+      "grad_norm": 0.4731711629332615,
+      "learning_rate": 5.740262745928293e-05,
+      "loss": 0.5448,
+      "step": 240
+    },
+    {
+      "epoch": 2.103598691384951,
+      "grad_norm": 0.5087479759223862,
+      "learning_rate": 5.718174220296949e-05,
+      "loss": 0.5415,
+      "step": 241
+    },
+    {
+      "epoch": 2.1123227917121046,
+      "grad_norm": 0.3397004171608698,
+      "learning_rate": 5.6960212583267873e-05,
+      "loss": 0.5373,
+      "step": 242
+    },
+    {
+      "epoch": 2.1210468920392582,
+      "grad_norm": 0.3028591738754104,
+      "learning_rate": 5.673804690815845e-05,
+      "loss": 0.5393,
+      "step": 243
+    },
+    {
+      "epoch": 2.1297709923664123,
+      "grad_norm": 0.4035449728515965,
+      "learning_rate": 5.6515253509475405e-05,
+      "loss": 0.5351,
+      "step": 244
+    },
+    {
+      "epoch": 2.138495092693566,
+      "grad_norm": 0.3353749295009557,
+      "learning_rate": 5.6291840742594305e-05,
+      "loss": 0.5365,
+      "step": 245
+    },
+    {
+      "epoch": 2.1472191930207196,
+      "grad_norm": 0.2948957119809547,
+      "learning_rate": 5.606781698611879e-05,
+      "loss": 0.5367,
+      "step": 246
+    },
+    {
+      "epoch": 2.1559432933478737,
+      "grad_norm": 0.3274847500759932,
+      "learning_rate": 5.584319064156628e-05,
+      "loss": 0.5343,
+      "step": 247
+    },
+    {
+      "epoch": 2.1646673936750274,
+      "grad_norm": 0.27942787362736154,
+      "learning_rate": 5.561797013305297e-05,
+      "loss": 0.5355,
+      "step": 248
+    },
+    {
+      "epoch": 2.173391494002181,
+      "grad_norm": 0.3320815797761542,
+      "learning_rate": 5.5392163906977835e-05,
+      "loss": 0.5324,
+      "step": 249
+    },
+    {
+      "epoch": 2.1821155943293347,
+      "grad_norm": 0.2793266605014413,
+      "learning_rate": 5.516578043170591e-05,
+      "loss": 0.5333,
+      "step": 250
+    },
+    {
+      "epoch": 2.1908396946564888,
+      "grad_norm": 0.24153435404173842,
+      "learning_rate": 5.49388281972507e-05,
+      "loss": 0.5403,
+      "step": 251
+    },
+    {
+      "epoch": 2.1995637949836424,
+      "grad_norm": 0.29337585767480123,
+      "learning_rate": 5.471131571495574e-05,
+      "loss": 0.5479,
+      "step": 252
+    },
+    {
+      "epoch": 2.208287895310796,
+      "grad_norm": 0.3065886341708783,
+      "learning_rate": 5.4483251517175454e-05,
+      "loss": 0.5274,
+      "step": 253
+    },
+    {
+      "epoch": 2.2170119956379497,
+      "grad_norm": 0.2885822613470245,
+      "learning_rate": 5.425464415695514e-05,
+      "loss": 0.5431,
+      "step": 254
+    },
+    {
+      "epoch": 2.225736095965104,
+      "grad_norm": 0.2588491126601207,
+      "learning_rate": 5.4025502207710184e-05,
+      "loss": 0.5336,
+      "step": 255
+    },
+    {
+      "epoch": 2.2344601962922575,
+      "grad_norm": 0.26723660749591893,
+      "learning_rate": 5.379583426290458e-05,
+      "loss": 0.5338,
+      "step": 256
+    },
+    {
+      "epoch": 2.243184296619411,
+      "grad_norm": 0.3305889777035953,
+      "learning_rate": 5.356564893572859e-05,
+      "loss": 0.5393,
+      "step": 257
+    },
+    {
+      "epoch": 2.2519083969465647,
+      "grad_norm": 0.22453022727431107,
+      "learning_rate": 5.333495485877583e-05,
+      "loss": 0.5485,
+      "step": 258
+    },
+    {
+      "epoch": 2.2606324972737184,
+      "grad_norm": 0.19233156617492056,
+      "learning_rate": 5.310376068371938e-05,
+      "loss": 0.5418,
+      "step": 259
+    },
+    {
+      "epoch": 2.2693565976008725,
+      "grad_norm": 0.20527709672078884,
+      "learning_rate": 5.287207508098743e-05,
+      "loss": 0.5416,
+      "step": 260
+    },
+    {
+      "epoch": 2.278080697928026,
+      "grad_norm": 0.2376527665778272,
+      "learning_rate": 5.263990673943811e-05,
+      "loss": 0.54,
+      "step": 261
+    },
+    {
+      "epoch": 2.28680479825518,
+      "grad_norm": 0.2613858051995676,
+      "learning_rate": 5.2407264366033555e-05,
+      "loss": 0.5498,
+      "step": 262
+    },
+    {
+      "epoch": 2.295528898582334,
+      "grad_norm": 0.18916730210716454,
+      "learning_rate": 5.2174156685513446e-05,
+      "loss": 0.5364,
+      "step": 263
+    },
+    {
+      "epoch": 2.3042529989094875,
+      "grad_norm": 0.19275577825236526,
+      "learning_rate": 5.194059244006779e-05,
+      "loss": 0.5434,
+      "step": 264
+    },
+    {
+      "epoch": 2.312977099236641,
+      "grad_norm": 0.2384097782515164,
+      "learning_rate": 5.170658038900904e-05,
+      "loss": 0.5353,
+      "step": 265
+    },
+    {
+      "epoch": 2.321701199563795,
+      "grad_norm": 0.21432203996579718,
+      "learning_rate": 5.1472129308443616e-05,
+      "loss": 0.537,
+      "step": 266
+    },
+    {
+      "epoch": 2.330425299890949,
+      "grad_norm": 0.17783572065614472,
+      "learning_rate": 5.123724799094279e-05,
+      "loss": 0.5444,
+      "step": 267
+    },
+    {
+      "epoch": 2.3391494002181026,
+      "grad_norm": 0.24973497434785122,
+      "learning_rate": 5.1001945245212874e-05,
+      "loss": 0.5404,
+      "step": 268
+    },
+    {
+      "epoch": 2.3478735005452562,
+      "grad_norm": 0.2176404792941727,
+      "learning_rate": 5.076622989576498e-05,
+      "loss": 0.5376,
+      "step": 269
+    },
+    {
+      "epoch": 2.35659760087241,
+      "grad_norm": 0.2460399652714288,
+      "learning_rate": 5.053011078258397e-05,
+      "loss": 0.5412,
+      "step": 270
+    },
+    {
+      "epoch": 2.365321701199564,
+      "grad_norm": 0.1827234799837285,
+      "learning_rate": 5.0293596760797e-05,
+      "loss": 0.5394,
+      "step": 271
+    },
+    {
+      "epoch": 2.3740458015267176,
+      "grad_norm": 0.21898045121125942,
+      "learning_rate": 5.005669670034138e-05,
+      "loss": 0.5417,
+      "step": 272
+    },
+    {
+      "epoch": 2.3827699018538713,
+      "grad_norm": 0.24065391523216037,
+      "learning_rate": 4.981941948563197e-05,
+      "loss": 0.5425,
+      "step": 273
+    },
+    {
+      "epoch": 2.391494002181025,
+      "grad_norm": 0.19964448205882962,
+      "learning_rate": 4.958177401522796e-05,
+      "loss": 0.5383,
+      "step": 274
+    },
+    {
+      "epoch": 2.400218102508179,
+      "grad_norm": 0.2287719342097617,
+      "learning_rate": 4.934376920149915e-05,
+      "loss": 0.5424,
+      "step": 275
+    },
+    {
+      "epoch": 2.4089422028353327,
+      "grad_norm": 0.15994920899181853,
+      "learning_rate": 4.9105413970291747e-05,
+      "loss": 0.5492,
+      "step": 276
+    },
+    {
+      "epoch": 2.4176663031624863,
+      "grad_norm": 0.20177344996011212,
+      "learning_rate": 4.886671726059355e-05,
+      "loss": 0.5452,
+      "step": 277
+    },
+    {
+      "epoch": 2.42639040348964,
+      "grad_norm": 0.2505950233818444,
+      "learning_rate": 4.862768802419881e-05,
+      "loss": 0.5355,
+      "step": 278
+    },
+    {
+      "epoch": 2.435114503816794,
+      "grad_norm": 0.20877969786025213,
+      "learning_rate": 4.8388335225372416e-05,
+      "loss": 0.5312,
+      "step": 279
+    },
+    {
+      "epoch": 2.4438386041439477,
+      "grad_norm": 0.16740200168268377,
+      "learning_rate": 4.8148667840513773e-05,
+      "loss": 0.536,
+      "step": 280
+    },
+    {
+      "epoch": 2.4525627044711014,
+      "grad_norm": 0.1624204063771831,
+      "learning_rate": 4.790869485782014e-05,
+      "loss": 0.5411,
+      "step": 281
+    },
+    {
+      "epoch": 2.461286804798255,
+      "grad_norm": 0.17548255884293712,
+      "learning_rate": 4.7668425276949546e-05,
+      "loss": 0.5377,
+      "step": 282
+    },
+    {
+      "epoch": 2.470010905125409,
+      "grad_norm": 0.24425295770194558,
+      "learning_rate": 4.742786810868327e-05,
+      "loss": 0.5424,
+      "step": 283
+    },
+    {
+      "epoch": 2.4787350054525628,
+      "grad_norm": 0.2279410111439652,
+      "learning_rate": 4.7187032374587956e-05,
+      "loss": 0.5408,
+      "step": 284
+    },
+    {
+      "epoch": 2.4874591057797164,
+      "grad_norm": 0.18333960694349968,
+      "learning_rate": 4.694592710667723e-05,
+      "loss": 0.5387,
+      "step": 285
+    },
+    {
+      "epoch": 2.49618320610687,
+      "grad_norm": 0.17387901999937752,
+      "learning_rate": 4.670456134707294e-05,
+      "loss": 0.5412,
+      "step": 286
+    },
+    {
+      "epoch": 2.504907306434024,
+      "grad_norm": 0.1802799576438455,
+      "learning_rate": 4.64629441476662e-05,
+      "loss": 0.5358,
+      "step": 287
+    },
+    {
+      "epoch": 2.513631406761178,
+      "grad_norm": 0.174702677057685,
+      "learning_rate": 4.622108456977773e-05,
+      "loss": 0.536,
+      "step": 288
+    },
+    {
+      "epoch": 2.5223555070883314,
+      "grad_norm": 0.14211263717330275,
+      "learning_rate": 4.597899168381818e-05,
+      "loss": 0.5441,
+      "step": 289
+    },
+    {
+      "epoch": 2.531079607415485,
+      "grad_norm": 0.1774683759788631,
+      "learning_rate": 4.573667456894786e-05,
+      "loss": 0.5349,
+      "step": 290
+    },
+    {
+      "epoch": 2.539803707742639,
+      "grad_norm": 0.1951727957490502,
+      "learning_rate": 4.549414231273633e-05,
+      "loss": 0.5428,
+      "step": 291
+    },
+    {
+      "epoch": 2.548527808069793,
+      "grad_norm": 0.16687560601341936,
+      "learning_rate": 4.525140401082153e-05,
+      "loss": 0.535,
+      "step": 292
+    },
+    {
+      "epoch": 2.5572519083969465,
+      "grad_norm": 0.15616561471578494,
+      "learning_rate": 4.50084687665687e-05,
+      "loss": 0.5475,
+      "step": 293
+    },
+    {
+      "epoch": 2.5659760087241006,
+      "grad_norm": 0.15824726116167803,
+      "learning_rate": 4.476534569072895e-05,
+      "loss": 0.5473,
+      "step": 294
+    },
+    {
+      "epoch": 2.5747001090512542,
+      "grad_norm": 0.22356058463377537,
+      "learning_rate": 4.452204390109763e-05,
+      "loss": 0.5345,
+      "step": 295
+    },
+    {
+      "epoch": 2.583424209378408,
+      "grad_norm": 0.21936785429021016,
+      "learning_rate": 4.4278572522172336e-05,
+      "loss": 0.5379,
+      "step": 296
+    },
+    {
+      "epoch": 2.5921483097055615,
+      "grad_norm": 0.15224856301200368,
+      "learning_rate": 4.403494068481074e-05,
+      "loss": 0.5409,
+      "step": 297
+    },
+    {
+      "epoch": 2.600872410032715,
+      "grad_norm": 0.19196426118514048,
+      "learning_rate": 4.379115752588814e-05,
+      "loss": 0.5531,
+      "step": 298
+    },
+    {
+      "epoch": 2.6095965103598693,
+      "grad_norm": 0.21896927123986717,
+      "learning_rate": 4.3547232187954866e-05,
+      "loss": 0.5437,
+      "step": 299
+    },
+    {
+      "epoch": 2.618320610687023,
+      "grad_norm": 0.187029673883834,
+      "learning_rate": 4.33031738188933e-05,
+      "loss": 0.5381,
+      "step": 300
+    },
+    {
+      "epoch": 2.6270447110141766,
+      "grad_norm": 0.1542884550374639,
+      "learning_rate": 4.3058991571574896e-05,
+      "loss": 0.5421,
+      "step": 301
+    },
+    {
+      "epoch": 2.6357688113413307,
+      "grad_norm": 0.2086016114061771,
+      "learning_rate": 4.2814694603516876e-05,
+      "loss": 0.5492,
+      "step": 302
+    },
+    {
+      "epoch": 2.6444929116684843,
+      "grad_norm": 0.19799737279991564,
+      "learning_rate": 4.257029207653881e-05,
+      "loss": 0.5372,
+      "step": 303
+    },
+    {
+      "epoch": 2.653217011995638,
+      "grad_norm": 0.16486499555492864,
+      "learning_rate": 4.2325793156419035e-05,
+      "loss": 0.5468,
+      "step": 304
+    },
+    {
+      "epoch": 2.6619411123227916,
+      "grad_norm": 0.21172626026429878,
+      "learning_rate": 4.20812070125509e-05,
+      "loss": 0.5396,
+      "step": 305
+    },
+    {
+      "epoch": 2.6706652126499453,
+      "grad_norm": 0.2470196212926822,
+      "learning_rate": 4.183654281759888e-05,
+      "loss": 0.5454,
+      "step": 306
+    },
+    {
+      "epoch": 2.6793893129770994,
+      "grad_norm": 0.22097827298910216,
+      "learning_rate": 4.159180974715457e-05,
+      "loss": 0.5401,
+      "step": 307
+    },
+    {
+      "epoch": 2.688113413304253,
+      "grad_norm": 0.15719194953902701,
+      "learning_rate": 4.1347016979392626e-05,
+      "loss": 0.5351,
+      "step": 308
+    },
+    {
+      "epoch": 2.6968375136314067,
+      "grad_norm": 0.15729742427040297,
+      "learning_rate": 4.110217369472649e-05,
+      "loss": 0.5295,
+      "step": 309
+    },
+    {
+      "epoch": 2.7055616139585608,
+      "grad_norm": 0.14739912075610104,
+      "learning_rate": 4.085728907546413e-05,
+      "loss": 0.5408,
+      "step": 310
+    },
+    {
+      "epoch": 2.7142857142857144,
+      "grad_norm": 0.14793685021802994,
+      "learning_rate": 4.061237230546369e-05,
+      "loss": 0.5428,
+      "step": 311
+    },
+    {
+      "epoch": 2.723009814612868,
+      "grad_norm": 0.16234675353976102,
+      "learning_rate": 4.0367432569789065e-05,
+      "loss": 0.5362,
+      "step": 312
+    },
+    {
+      "epoch": 2.7317339149400217,
+      "grad_norm": 0.14302498512172665,
+      "learning_rate": 4.012247905436539e-05,
+      "loss": 0.5412,
+      "step": 313
+    },
+    {
+      "epoch": 2.7404580152671754,
+      "grad_norm": 0.14016241879849234,
+      "learning_rate": 3.987752094563462e-05,
+      "loss": 0.5396,
+      "step": 314
+    },
+    {
+      "epoch": 2.7491821155943295,
+      "grad_norm": 0.15182946842699183,
+      "learning_rate": 3.963256743021095e-05,
+      "loss": 0.54,
+      "step": 315
+    },
+    {
+      "epoch": 2.757906215921483,
+      "grad_norm": 0.14224077108756453,
+      "learning_rate": 3.9387627694536316e-05,
+      "loss": 0.541,
+      "step": 316
+    },
+    {
+      "epoch": 2.7666303162486368,
+      "grad_norm": 0.15804186928200673,
+      "learning_rate": 3.914271092453589e-05,
+      "loss": 0.5436,
+      "step": 317
+    },
+    {
+      "epoch": 2.775354416575791,
+      "grad_norm": 0.18867349297613747,
+      "learning_rate": 3.889782630527353e-05,
+      "loss": 0.5303,
+      "step": 318
+    },
+    {
+      "epoch": 2.7840785169029445,
+      "grad_norm": 0.14326765308764183,
+      "learning_rate": 3.865298302060739e-05,
+      "loss": 0.5306,
+      "step": 319
+    },
+    {
+      "epoch": 2.792802617230098,
+      "grad_norm": 0.2074238009547974,
+      "learning_rate": 3.8408190252845435e-05,
+      "loss": 0.5372,
+      "step": 320
+    },
+    {
+      "epoch": 2.801526717557252,
+      "grad_norm": 0.14034785022531543,
+      "learning_rate": 3.816345718240113e-05,
+      "loss": 0.5403,
+      "step": 321
+    },
+    {
+      "epoch": 2.8102508178844054,
+      "grad_norm": 0.15821673087787588,
+      "learning_rate": 3.791879298744911e-05,
+      "loss": 0.5321,
+      "step": 322
+    },
+    {
+      "epoch": 2.8189749182115595,
+      "grad_norm": 0.14024264925437085,
+      "learning_rate": 3.767420684358097e-05,
+      "loss": 0.5509,
+      "step": 323
+    },
+    {
+      "epoch": 2.827699018538713,
+      "grad_norm": 0.14817524358908055,
+      "learning_rate": 3.74297079234612e-05,
+      "loss": 0.5414,
+      "step": 324
+    },
+    {
+      "epoch": 2.836423118865867,
+      "grad_norm": 0.1618059561441321,
+      "learning_rate": 3.7185305396483144e-05,
+      "loss": 0.5435,
+      "step": 325
+    },
+    {
+      "epoch": 2.845147219193021,
+      "grad_norm": 0.1463097964083179,
+      "learning_rate": 3.694100842842512e-05,
+      "loss": 0.5415,
+      "step": 326
+    },
+    {
+      "epoch": 2.8538713195201746,
+      "grad_norm": 0.19816090369688377,
+      "learning_rate": 3.669682618110671e-05,
+      "loss": 0.5289,
+      "step": 327
+    },
+    {
+      "epoch": 2.8625954198473282,
+      "grad_norm": 0.13008063298224137,
+      "learning_rate": 3.645276781204515e-05,
+      "loss": 0.5409,
+      "step": 328
+    },
+    {
+      "epoch": 2.871319520174482,
+      "grad_norm": 0.19611647889519343,
+      "learning_rate": 3.6208842474111865e-05,
+      "loss": 0.5318,
+      "step": 329
+    },
+    {
+      "epoch": 2.8800436205016355,
+      "grad_norm": 0.16795674711289083,
+      "learning_rate": 3.5965059315189274e-05,
+      "loss": 0.5402,
+      "step": 330
+    },
+    {
+      "epoch": 2.8887677208287896,
+      "grad_norm": 0.15014005665113414,
+      "learning_rate": 3.572142747782768e-05,
+      "loss": 0.5358,
+      "step": 331
+    },
+    {
+      "epoch": 2.8974918211559433,
+      "grad_norm": 0.17800420417158852,
+      "learning_rate": 3.547795609890238e-05,
+      "loss": 0.5434,
+      "step": 332
+    },
+    {
+      "epoch": 2.906215921483097,
+      "grad_norm": 0.15795311662461162,
+      "learning_rate": 3.523465430927106e-05,
+      "loss": 0.5391,
+      "step": 333
+    },
+    {
+      "epoch": 2.914940021810251,
+      "grad_norm": 0.12743060334259004,
+      "learning_rate": 3.499153123343131e-05,
+      "loss": 0.5325,
+      "step": 334
+    },
+    {
+      "epoch": 2.9236641221374047,
+      "grad_norm": 0.16855028063385719,
+      "learning_rate": 3.474859598917849e-05,
+      "loss": 0.5413,
+      "step": 335
+    },
+    {
+      "epoch": 2.9323882224645583,
+      "grad_norm": 0.13048706187884607,
+      "learning_rate": 3.4505857687263675e-05,
+      "loss": 0.5424,
+      "step": 336
+    },
+    {
+      "epoch": 2.941112322791712,
+      "grad_norm": 0.19367317061005948,
+      "learning_rate": 3.4263325431052156e-05,
+      "loss": 0.5432,
+      "step": 337
+    },
+    {
+      "epoch": 2.9498364231188656,
+      "grad_norm": 0.12027268336619701,
+      "learning_rate": 3.4021008316181834e-05,
+      "loss": 0.5454,
+      "step": 338
+    },
+    {
+      "epoch": 2.9585605234460197,
+      "grad_norm": 0.15147251810786375,
+      "learning_rate": 3.377891543022229e-05,
+      "loss": 0.5319,
+      "step": 339
+    },
+    {
+      "epoch": 2.9672846237731734,
+      "grad_norm": 0.12811996405452336,
+      "learning_rate": 3.353705585233381e-05,
+      "loss": 0.544,
+      "step": 340
+    },
+    {
+      "epoch": 2.976008724100327,
+      "grad_norm": 0.13491661704715102,
+      "learning_rate": 3.329543865292707e-05,
+      "loss": 0.5435,
+      "step": 341
+    },
+    {
+      "epoch": 2.984732824427481,
+      "grad_norm": 0.15940079876011493,
+      "learning_rate": 3.305407289332279e-05,
+      "loss": 0.5321,
+      "step": 342
+    },
+    {
+      "epoch": 2.9934569247546348,
+      "grad_norm": 0.13416535357448142,
+      "learning_rate": 3.281296762541206e-05,
+      "loss": 0.5555,
+      "step": 343
+    },
+    {
+      "epoch": 3.0174482006543077,
+      "grad_norm": 0.2721337362560336,
+      "learning_rate": 3.2572131891316735e-05,
+      "loss": 0.4954,
+      "step": 344
+    },
+    {
+      "epoch": 3.0261723009814614,
+      "grad_norm": 0.19428239960148935,
+      "learning_rate": 3.2331574723050474e-05,
+      "loss": 0.4841,
+      "step": 345
+    },
+    {
+      "epoch": 3.034896401308615,
+      "grad_norm": 0.21517719571854402,
+      "learning_rate": 3.209130514217987e-05,
+      "loss": 0.4868,
+      "step": 346
+    },
+    {
+      "epoch": 3.0436205016357687,
+      "grad_norm": 0.21889099316423938,
+      "learning_rate": 3.185133215948625e-05,
+      "loss": 0.4986,
+      "step": 347
+    },
+    {
+      "epoch": 3.052344601962923,
+      "grad_norm": 0.21703320950794194,
+      "learning_rate": 3.161166477462759e-05,
+      "loss": 0.4879,
+      "step": 348
+    },
+    {
+      "epoch": 3.0610687022900764,
+      "grad_norm": 0.20716597340235016,
+      "learning_rate": 3.137231197580121e-05,
+      "loss": 0.4926,
+      "step": 349
+    },
+    {
+      "epoch": 3.06979280261723,
+      "grad_norm": 0.23215045785693364,
+      "learning_rate": 3.1133282739406454e-05,
+      "loss": 0.4823,
+      "step": 350
+    },
+    {
+      "epoch": 3.0785169029443837,
+      "grad_norm": 0.18540346216841436,
+      "learning_rate": 3.089458602970828e-05,
+      "loss": 0.4932,
+      "step": 351
+    },
+    {
+      "epoch": 3.087241003271538,
+      "grad_norm": 0.2117646196498009,
+      "learning_rate": 3.065623079850086e-05,
+      "loss": 0.4829,
+      "step": 352
+    },
+    {
+      "epoch": 3.0959651035986915,
+      "grad_norm": 0.19784374074464264,
+      "learning_rate": 3.041822598477206e-05,
+      "loss": 0.4894,
+      "step": 353
+    },
+    {
+      "epoch": 3.104689203925845,
+      "grad_norm": 0.20389277832289257,
+      "learning_rate": 3.0180580514368037e-05,
+      "loss": 0.4819,
+      "step": 354
+    },
+    {
+      "epoch": 3.113413304252999,
+      "grad_norm": 0.20689059795484877,
+      "learning_rate": 2.9943303299658634e-05,
+      "loss": 0.4854,
+      "step": 355
+    },
+    {
+      "epoch": 3.122137404580153,
+      "grad_norm": 0.19252775326832058,
+      "learning_rate": 2.970640323920301e-05,
+      "loss": 0.4853,
+      "step": 356
+    },
+    {
+      "epoch": 3.1308615049073065,
+      "grad_norm": 0.20405962502922226,
+      "learning_rate": 2.9469889217416045e-05,
+      "loss": 0.4931,
+      "step": 357
+    },
+    {
+      "epoch": 3.13958560523446,
+      "grad_norm": 0.18417287655063955,
+      "learning_rate": 2.9233770104235027e-05,
+      "loss": 0.4861,
+      "step": 358
+    },
+    {
+      "epoch": 3.148309705561614,
+      "grad_norm": 0.17779606526419078,
+      "learning_rate": 2.8998054754787143e-05,
+      "loss": 0.4892,
+      "step": 359
+    },
+    {
+      "epoch": 3.157033805888768,
+      "grad_norm": 0.17500936627057292,
+      "learning_rate": 2.8762752009057232e-05,
+      "loss": 0.482,
+      "step": 360
+    },
+    {
+      "epoch": 3.1657579062159216,
+      "grad_norm": 0.14921104868796375,
+      "learning_rate": 2.8527870691556404e-05,
+      "loss": 0.4884,
+      "step": 361
+    },
+    {
+      "epoch": 3.174482006543075,
+      "grad_norm": 0.1738429047868548,
+      "learning_rate": 2.8293419610990968e-05,
+      "loss": 0.4856,
+      "step": 362
+    },
+    {
+      "epoch": 3.183206106870229,
+      "grad_norm": 0.14815153041649903,
+      "learning_rate": 2.805940755993223e-05,
+      "loss": 0.4856,
+      "step": 363
+    },
+    {
+      "epoch": 3.191930207197383,
+      "grad_norm": 0.16213628899803448,
+      "learning_rate": 2.7825843314486564e-05,
+      "loss": 0.4833,
+      "step": 364
+    },
+    {
+      "epoch": 3.2006543075245366,
+      "grad_norm": 0.12776465267608372,
+      "learning_rate": 2.7592735633966468e-05,
+      "loss": 0.4825,
+      "step": 365
+    },
+    {
+      "epoch": 3.2093784078516903,
+      "grad_norm": 0.1532685309345078,
+      "learning_rate": 2.7360093260561904e-05,
+      "loss": 0.4865,
+      "step": 366
+    },
+    {
+      "epoch": 3.218102508178844,
+      "grad_norm": 0.13916327010405047,
+      "learning_rate": 2.712792491901258e-05,
+      "loss": 0.4784,
+      "step": 367
+    },
+    {
+      "epoch": 3.226826608505998,
+      "grad_norm": 0.13219062695260514,
+      "learning_rate": 2.6896239316280634e-05,
+      "loss": 0.4883,
+      "step": 368
+    },
+    {
+      "epoch": 3.2355507088331517,
+      "grad_norm": 0.1425947319673384,
+      "learning_rate": 2.6665045141224193e-05,
+      "loss": 0.4842,
+      "step": 369
+    },
+    {
+      "epoch": 3.2442748091603053,
+      "grad_norm": 0.13401395028938867,
+      "learning_rate": 2.643435106427141e-05,
+      "loss": 0.4911,
+      "step": 370
+    },
+    {
+      "epoch": 3.252998909487459,
+      "grad_norm": 0.13526534167394852,
+      "learning_rate": 2.6204165737095447e-05,
+      "loss": 0.486,
+      "step": 371
+    },
+    {
+      "epoch": 3.261723009814613,
+      "grad_norm": 0.13067389909619231,
+      "learning_rate": 2.597449779228983e-05,
+      "loss": 0.4915,
+      "step": 372
+    },
+    {
+      "epoch": 3.2704471101417667,
+      "grad_norm": 0.12011326204522878,
+      "learning_rate": 2.5745355843044882e-05,
+      "loss": 0.4863,
+      "step": 373
+    },
+    {
+      "epoch": 3.2791712104689203,
+      "grad_norm": 0.12879501353541944,
+      "learning_rate": 2.5516748482824552e-05,
+      "loss": 0.482,
+      "step": 374
+    },
+    {
+      "epoch": 3.287895310796074,
+      "grad_norm": 0.13720739097629261,
+      "learning_rate": 2.5288684285044283e-05,
+      "loss": 0.4949,
+      "step": 375
+    },
+    {
+      "epoch": 3.296619411123228,
+      "grad_norm": 0.1302222305392029,
+      "learning_rate": 2.5061171802749317e-05,
+      "loss": 0.494,
+      "step": 376
+    },
+    {
+      "epoch": 3.3053435114503817,
+      "grad_norm": 0.13160648955568055,
+      "learning_rate": 2.4834219568294112e-05,
+      "loss": 0.4813,
+      "step": 377
+    },
+    {
+      "epoch": 3.3140676117775354,
+      "grad_norm": 0.12345223103133267,
+      "learning_rate": 2.460783609302218e-05,
+      "loss": 0.4872,
+      "step": 378
+    },
+    {
+      "epoch": 3.322791712104689,
+      "grad_norm": 0.12449345441651119,
+      "learning_rate": 2.438202986694705e-05,
+      "loss": 0.4779,
+      "step": 379
+    },
+    {
+      "epoch": 3.331515812431843,
+      "grad_norm": 0.13788217467466876,
+      "learning_rate": 2.4156809358433728e-05,
+      "loss": 0.4827,
+      "step": 380
+    },
+    {
+      "epoch": 3.340239912758997,
+      "grad_norm": 0.13887142392886914,
+      "learning_rate": 2.393218301388123e-05,
+      "loss": 0.4805,
+      "step": 381
+    },
+    {
+      "epoch": 3.3489640130861504,
+      "grad_norm": 0.1279136541009379,
+      "learning_rate": 2.3708159257405705e-05,
+      "loss": 0.4833,
+      "step": 382
+    },
+    {
+      "epoch": 3.357688113413304,
+      "grad_norm": 0.12043380955651341,
+      "learning_rate": 2.348474649052462e-05,
+      "loss": 0.4864,
+      "step": 383
+    },
+    {
+      "epoch": 3.366412213740458,
+      "grad_norm": 0.12056545888603093,
+      "learning_rate": 2.3261953091841553e-05,
+      "loss": 0.4946,
+      "step": 384
+    },
+    {
+      "epoch": 3.375136314067612,
+      "grad_norm": 0.12834970766272358,
+      "learning_rate": 2.3039787416732143e-05,
+      "loss": 0.4883,
+      "step": 385
+    },
+    {
+      "epoch": 3.3838604143947655,
+      "grad_norm": 0.12525269244211057,
+      "learning_rate": 2.281825779703054e-05,
+      "loss": 0.4848,
+      "step": 386
+    },
+    {
+      "epoch": 3.392584514721919,
+      "grad_norm": 0.11069140829422457,
+      "learning_rate": 2.2597372540717083e-05,
+      "loss": 0.49,
+      "step": 387
+    },
+    {
+      "epoch": 3.4013086150490732,
+      "grad_norm": 0.1186817635248957,
+      "learning_rate": 2.237713993160668e-05,
+      "loss": 0.4904,
+      "step": 388
+    },
+    {
+      "epoch": 3.410032715376227,
+      "grad_norm": 0.11808550288276015,
+      "learning_rate": 2.2157568229038145e-05,
+      "loss": 0.4902,
+      "step": 389
+    },
+    {
+      "epoch": 3.4187568157033805,
+      "grad_norm": 0.11993464935882125,
+      "learning_rate": 2.1938665667564435e-05,
+      "loss": 0.4845,
+      "step": 390
+    },
+    {
+      "epoch": 3.427480916030534,
+      "grad_norm": 0.11969074606441606,
+      "learning_rate": 2.172044045664386e-05,
+      "loss": 0.5019,
+      "step": 391
+    },
+    {
+      "epoch": 3.4362050163576883,
+      "grad_norm": 0.12038618305452632,
+      "learning_rate": 2.1502900780332183e-05,
+      "loss": 0.4901,
+      "step": 392
+    },
+    {
+      "epoch": 3.444929116684842,
+      "grad_norm": 0.11405277423193325,
+      "learning_rate": 2.1286054796975696e-05,
+      "loss": 0.4869,
+      "step": 393
+    },
+    {
+      "epoch": 3.4536532170119956,
+      "grad_norm": 0.13087116805331267,
+      "learning_rate": 2.1069910638905277e-05,
+      "loss": 0.4894,
+      "step": 394
+    },
+    {
+      "epoch": 3.462377317339149,
+      "grad_norm": 0.10658227722079491,
+      "learning_rate": 2.085447641213138e-05,
+      "loss": 0.4873,
+      "step": 395
+    },
+    {
+      "epoch": 3.4711014176663033,
+      "grad_norm": 0.11425468095535157,
+      "learning_rate": 2.063976019604006e-05,
+      "loss": 0.4831,
+      "step": 396
+    },
+    {
+      "epoch": 3.479825517993457,
+      "grad_norm": 0.11858217904877387,
+      "learning_rate": 2.0425770043089957e-05,
+      "loss": 0.4825,
+      "step": 397
+    },
+    {
+      "epoch": 3.4885496183206106,
+      "grad_norm": 0.10999883286826867,
+      "learning_rate": 2.021251397851031e-05,
+      "loss": 0.485,
+      "step": 398
+    },
+    {
+      "epoch": 3.4972737186477643,
+      "grad_norm": 0.11453238535992676,
+      "learning_rate": 2.0000000000000012e-05,
+      "loss": 0.4771,
+      "step": 399
+    },
+    {
+      "epoch": 3.5059978189749184,
+      "grad_norm": 0.11260955156023703,
+      "learning_rate": 1.9788236077427617e-05,
+      "loss": 0.4849,
+      "step": 400
+    },
+    {
+      "epoch": 3.514721919302072,
+      "grad_norm": 0.10788933054700972,
+      "learning_rate": 1.957723015253252e-05,
+      "loss": 0.4879,
+      "step": 401
+    },
+    {
+      "epoch": 3.5234460196292257,
+      "grad_norm": 0.1242817430775905,
+      "learning_rate": 1.9366990138627054e-05,
+      "loss": 0.4954,
+      "step": 402
+    },
+    {
+      "epoch": 3.5321701199563798,
+      "grad_norm": 0.1250660403349655,
+      "learning_rate": 1.915752392029976e-05,
+      "loss": 0.4873,
+      "step": 403
+    },
+    {
+      "epoch": 3.5408942202835334,
+      "grad_norm": 0.10866091824566891,
+      "learning_rate": 1.894883935311969e-05,
+      "loss": 0.4938,
+      "step": 404
+    },
+    {
+      "epoch": 3.549618320610687,
+      "grad_norm": 0.134004404392773,
+      "learning_rate": 1.8740944263341773e-05,
+      "loss": 0.4842,
+      "step": 405
+    },
+    {
+      "epoch": 3.5583424209378407,
+      "grad_norm": 0.10798242739035056,
+      "learning_rate": 1.8533846447613342e-05,
+      "loss": 0.489,
+      "step": 406
+    },
+    {
+      "epoch": 3.5670665212649943,
+      "grad_norm": 0.12508201458500512,
+      "learning_rate": 1.8327553672681717e-05,
+      "loss": 0.4887,
+      "step": 407
+    },
+    {
+      "epoch": 3.5757906215921484,
+      "grad_norm": 0.10902393852897485,
+      "learning_rate": 1.8122073675102935e-05,
+      "loss": 0.4794,
+      "step": 408
+    },
+    {
+      "epoch": 3.584514721919302,
+      "grad_norm": 0.1147667540376711,
+      "learning_rate": 1.7917414160951612e-05,
+      "loss": 0.4926,
+      "step": 409
+    },
+    {
+      "epoch": 3.5932388222464557,
+      "grad_norm": 0.10301491583219155,
+      "learning_rate": 1.7713582805531938e-05,
+      "loss": 0.4855,
+      "step": 410
+    },
+    {
+      "epoch": 3.60196292257361,
+      "grad_norm": 0.11784283632033268,
+      "learning_rate": 1.7510587253089842e-05,
+      "loss": 0.4923,
+      "step": 411
+    },
+    {
+      "epoch": 3.6106870229007635,
+      "grad_norm": 0.11268723502106516,
+      "learning_rate": 1.7308435116526293e-05,
+      "loss": 0.4798,
+      "step": 412
+    },
+    {
+      "epoch": 3.619411123227917,
+      "grad_norm": 0.10107252924229544,
+      "learning_rate": 1.7107133977111815e-05,
+      "loss": 0.4903,
+      "step": 413
+    },
+    {
+      "epoch": 3.628135223555071,
+      "grad_norm": 0.12328942904698265,
+      "learning_rate": 1.690669138420215e-05,
+      "loss": 0.4958,
+      "step": 414
+    },
+    {
+      "epoch": 3.6368593238822244,
+      "grad_norm": 0.10387729435616382,
+      "learning_rate": 1.6707114854955148e-05,
+      "loss": 0.4833,
+      "step": 415
+    },
+    {
+      "epoch": 3.6455834242093785,
+      "grad_norm": 0.10945876226281129,
+      "learning_rate": 1.6508411874048856e-05,
+      "loss": 0.4883,
+      "step": 416
+    },
+    {
+      "epoch": 3.654307524536532,
+      "grad_norm": 0.10647402251039108,
+      "learning_rate": 1.6310589893400804e-05,
+      "loss": 0.4888,
+      "step": 417
+    },
+    {
+      "epoch": 3.663031624863686,
+      "grad_norm": 0.10316209865500071,
+      "learning_rate": 1.6113656331888563e-05,
+      "loss": 0.4867,
+      "step": 418
+    },
+    {
+      "epoch": 3.67175572519084,
+      "grad_norm": 0.10970228886677176,
+      "learning_rate": 1.5917618575071482e-05,
+      "loss": 0.4805,
+      "step": 419
+    },
+    {
+      "epoch": 3.6804798255179936,
+      "grad_norm": 0.10554815817345704,
+      "learning_rate": 1.5722483974913737e-05,
+      "loss": 0.4895,
+      "step": 420
+    },
+    {
+      "epoch": 3.6892039258451472,
+      "grad_norm": 0.09884727405243657,
+      "learning_rate": 1.5528259849508617e-05,
+      "loss": 0.4869,
+      "step": 421
+    },
+    {
+      "epoch": 3.697928026172301,
+      "grad_norm": 0.11478692305275993,
+      "learning_rate": 1.533495348280404e-05,
+      "loss": 0.492,
+      "step": 422
+    },
+    {
+      "epoch": 3.7066521264994545,
+      "grad_norm": 0.10056955523850906,
+      "learning_rate": 1.5142572124329418e-05,
+      "loss": 0.4952,
+      "step": 423
+    },
+    {
+      "epoch": 3.7153762268266086,
+      "grad_norm": 0.10282422328881895,
+      "learning_rate": 1.4951122988923765e-05,
+      "loss": 0.4928,
+      "step": 424
+    },
+    {
+      "epoch": 3.7241003271537623,
+      "grad_norm": 0.11131425137941661,
+      "learning_rate": 1.4760613256465121e-05,
+      "loss": 0.4895,
+      "step": 425
+    },
+    {
+      "epoch": 3.732824427480916,
+      "grad_norm": 0.10147036335235084,
+      "learning_rate": 1.457105007160129e-05,
+      "loss": 0.4891,
+      "step": 426
+    },
+    {
+      "epoch": 3.74154852780807,
+      "grad_norm": 0.0979128656061041,
+      "learning_rate": 1.4382440543481888e-05,
+      "loss": 0.4962,
+      "step": 427
+    },
+    {
+      "epoch": 3.7502726281352237,
+      "grad_norm": 0.09836700247355973,
+      "learning_rate": 1.4194791745491748e-05,
+      "loss": 0.4887,
+      "step": 428
+    },
+    {
+      "epoch": 3.7589967284623773,
+      "grad_norm": 0.1025226008534582,
+      "learning_rate": 1.4008110714985623e-05,
+      "loss": 0.4921,
+      "step": 429
+    },
+    {
+      "epoch": 3.767720828789531,
+      "grad_norm": 0.09532630365560431,
+      "learning_rate": 1.3822404453024292e-05,
+      "loss": 0.4817,
+      "step": 430
+    },
+    {
+      "epoch": 3.7764449291166846,
+      "grad_norm": 0.10123230213257639,
+      "learning_rate": 1.363767992411197e-05,
+      "loss": 0.4902,
+      "step": 431
+    },
+    {
+      "epoch": 3.7851690294438387,
+      "grad_norm": 0.0991247900008933,
+      "learning_rate": 1.3453944055935151e-05,
+      "loss": 0.4848,
+      "step": 432
+    },
+    {
+      "epoch": 3.7938931297709924,
+      "grad_norm": 0.10006268884120084,
+      "learning_rate": 1.3271203739102783e-05,
+      "loss": 0.4964,
+      "step": 433
+    },
+    {
+      "epoch": 3.802617230098146,
+      "grad_norm": 0.10054874848417918,
+      "learning_rate": 1.308946582688785e-05,
+      "loss": 0.4922,
+      "step": 434
+    },
+    {
+      "epoch": 3.8113413304253,
+      "grad_norm": 0.0987785901332479,
+      "learning_rate": 1.2908737134970367e-05,
+      "loss": 0.4939,
+      "step": 435
+    },
+    {
+      "epoch": 3.8200654307524538,
+      "grad_norm": 0.09483259622573526,
+      "learning_rate": 1.272902444118175e-05,
+      "loss": 0.4844,
+      "step": 436
+    },
+    {
+      "epoch": 3.8287895310796074,
+      "grad_norm": 0.10027042176264714,
+      "learning_rate": 1.2550334485250661e-05,
+      "loss": 0.4961,
+      "step": 437
+    },
+    {
+      "epoch": 3.837513631406761,
+      "grad_norm": 0.10249666075304369,
+      "learning_rate": 1.2372673968550229e-05,
+      "loss": 0.485,
+      "step": 438
+    },
+    {
+      "epoch": 3.8462377317339147,
+      "grad_norm": 0.09908348278775064,
+      "learning_rate": 1.2196049553846718e-05,
+      "loss": 0.4929,
+      "step": 439
+    },
+    {
+      "epoch": 3.854961832061069,
+      "grad_norm": 0.10609152583672458,
+      "learning_rate": 1.202046786504969e-05,
+      "loss": 0.4899,
+      "step": 440
+    },
+    {
+      "epoch": 3.8636859323882224,
+      "grad_norm": 0.10563226786853933,
+      "learning_rate": 1.1845935486963546e-05,
+      "loss": 0.4897,
+      "step": 441
+    },
+    {
+      "epoch": 3.872410032715376,
+      "grad_norm": 0.09809013057751448,
+      "learning_rate": 1.1672458965040624e-05,
+      "loss": 0.484,
+      "step": 442
+    },
+    {
+      "epoch": 3.88113413304253,
+      "grad_norm": 0.10049787341345903,
+      "learning_rate": 1.1500044805135686e-05,
+      "loss": 0.4891,
+      "step": 443
+    },
+    {
+      "epoch": 3.889858233369684,
+      "grad_norm": 0.10177360915060568,
+      "learning_rate": 1.1328699473261957e-05,
+      "loss": 0.4894,
+      "step": 444
+    },
+    {
+      "epoch": 3.8985823336968375,
+      "grad_norm": 0.09437906218967344,
+      "learning_rate": 1.1158429395348614e-05,
+      "loss": 0.4883,
+      "step": 445
+    },
+    {
+      "epoch": 3.907306434023991,
+      "grad_norm": 0.10239261134633734,
+      "learning_rate": 1.0989240956999807e-05,
+      "loss": 0.4849,
+      "step": 446
+    },
+    {
+      "epoch": 3.916030534351145,
+      "grad_norm": 0.11040622306921051,
+      "learning_rate": 1.0821140503255174e-05,
+      "loss": 0.4856,
+      "step": 447
+    },
+    {
+      "epoch": 3.924754634678299,
+      "grad_norm": 0.10030020907072919,
+      "learning_rate": 1.0654134338351896e-05,
+      "loss": 0.4963,
+      "step": 448
+    },
+    {
+      "epoch": 3.9334787350054525,
+      "grad_norm": 0.10185833012036703,
+      "learning_rate": 1.0488228725488252e-05,
+      "loss": 0.4781,
+      "step": 449
+    },
+    {
+      "epoch": 3.942202835332606,
+      "grad_norm": 0.11302408247244193,
+      "learning_rate": 1.0323429886588743e-05,
+      "loss": 0.4952,
+      "step": 450
+    },
+    {
+      "epoch": 3.9509269356597603,
+      "grad_norm": 0.10091602155398324,
+      "learning_rate": 1.0159744002070756e-05,
+      "loss": 0.4843,
+      "step": 451
+    },
+    {
+      "epoch": 3.959651035986914,
+      "grad_norm": 0.0981028735736355,
+      "learning_rate": 9.997177210612774e-06,
+      "loss": 0.4705,
+      "step": 452
+    },
+    {
+      "epoch": 3.9683751363140676,
+      "grad_norm": 0.09982190269637335,
+      "learning_rate": 9.835735608924155e-06,
+      "loss": 0.4848,
+      "step": 453
+    },
+    {
+      "epoch": 3.9770992366412212,
+      "grad_norm": 0.10286988193952572,
+      "learning_rate": 9.675425251516502e-06,
+      "loss": 0.4862,
+      "step": 454
+    },
+    {
+      "epoch": 3.985823336968375,
+      "grad_norm": 0.09915491694077962,
+      "learning_rate": 9.516252150476583e-06,
+      "loss": 0.4822,
+      "step": 455
+    },
+    {
+      "epoch": 3.994547437295529,
+      "grad_norm": 0.1015074890493012,
+      "learning_rate": 9.358222275240884e-06,
+      "loss": 0.486,
+      "step": 456
+    },
+    {
+      "epoch": 4.003271537622683,
+      "grad_norm": 0.14625051321820182,
+      "learning_rate": 9.201341552371716e-06,
+      "loss": 0.6588,
+      "step": 457
+    },
+    {
+      "epoch": 4.011995637949837,
+      "grad_norm": 0.1728805026913459,
+      "learning_rate": 9.045615865334967e-06,
+      "loss": 0.4552,
+      "step": 458
+    },
+    {
+      "epoch": 4.02071973827699,
+      "grad_norm": 0.12344088598274187,
+      "learning_rate": 8.89105105427945e-06,
+      "loss": 0.4526,
+      "step": 459
+    },
+    {
+      "epoch": 4.029443838604144,
+      "grad_norm": 0.11496115531658294,
+      "learning_rate": 8.737652915817874e-06,
+      "loss": 0.4551,
+      "step": 460
+    },
+    {
+      "epoch": 4.038167938931298,
+      "grad_norm": 0.12574936341280377,
+      "learning_rate": 8.585427202809482e-06,
+      "loss": 0.4601,
+      "step": 461
+    },
+    {
+      "epoch": 4.046892039258451,
+      "grad_norm": 0.14155539924575067,
+      "learning_rate": 8.434379624144261e-06,
+      "loss": 0.4527,
+      "step": 462
+    },
+    {
+      "epoch": 4.055616139585605,
+      "grad_norm": 0.13406502403821957,
+      "learning_rate": 8.284515844528882e-06,
+      "loss": 0.455,
+      "step": 463
+    },
+    {
+      "epoch": 4.064340239912759,
+      "grad_norm": 0.13464345025289945,
+      "learning_rate": 8.13584148427423e-06,
+      "loss": 0.4543,
+      "step": 464
+    },
+    {
+      "epoch": 4.073064340239913,
+      "grad_norm": 0.12328299842024298,
+      "learning_rate": 7.988362119084642e-06,
+      "loss": 0.4515,
+      "step": 465
+    },
+    {
+      "epoch": 4.081788440567067,
+      "grad_norm": 0.11551422213866323,
+      "learning_rate": 7.842083279848797e-06,
+      "loss": 0.4592,
+      "step": 466
+    },
+    {
+      "epoch": 4.0905125408942205,
+      "grad_norm": 0.12057002567967279,
+      "learning_rate": 7.697010452432291e-06,
+      "loss": 0.4593,
+      "step": 467
+    },
+    {
+      "epoch": 4.099236641221374,
+      "grad_norm": 0.1289123182899191,
+      "learning_rate": 7.553149077471915e-06,
+      "loss": 0.457,
+      "step": 468
+    },
+    {
+      "epoch": 4.107960741548528,
+      "grad_norm": 0.11447570190961438,
+      "learning_rate": 7.410504550171582e-06,
+      "loss": 0.4523,
+      "step": 469
+    },
+    {
+      "epoch": 4.116684841875681,
+      "grad_norm": 0.1207957366113367,
+      "learning_rate": 7.269082220100037e-06,
+      "loss": 0.4607,
+      "step": 470
+    },
+    {
+      "epoch": 4.125408942202835,
+      "grad_norm": 0.1071935205957041,
+      "learning_rate": 7.128887390990198e-06,
+      "loss": 0.4488,
+      "step": 471
+    },
+    {
+      "epoch": 4.134133042529989,
+      "grad_norm": 0.11403879370654375,
+      "learning_rate": 6.989925320540267e-06,
+      "loss": 0.4528,
+      "step": 472
+    },
+    {
+      "epoch": 4.142857142857143,
+      "grad_norm": 0.11221608503491912,
+      "learning_rate": 6.852201220216547e-06,
+      "loss": 0.452,
+      "step": 473
+    },
+    {
+      "epoch": 4.151581243184297,
+      "grad_norm": 0.10102151712818268,
+      "learning_rate": 6.715720255058e-06,
+      "loss": 0.4461,
+      "step": 474
+    },
+    {
+      "epoch": 4.1603053435114505,
+      "grad_norm": 0.10159080301671279,
+      "learning_rate": 6.58048754348255e-06,
+      "loss": 0.4558,
+      "step": 475
+    },
+    {
+      "epoch": 4.169029443838604,
+      "grad_norm": 0.10300993558932357,
+      "learning_rate": 6.446508157095101e-06,
+      "loss": 0.4577,
+      "step": 476
+    },
+    {
+      "epoch": 4.177753544165758,
+      "grad_norm": 0.10298797711650444,
+      "learning_rate": 6.313787120497376e-06,
+      "loss": 0.4662,
+      "step": 477
+    },
+    {
+      "epoch": 4.1864776444929115,
+      "grad_norm": 0.10023066907847433,
+      "learning_rate": 6.18232941109945e-06,
+      "loss": 0.458,
+      "step": 478
+    },
+    {
+      "epoch": 4.195201744820065,
+      "grad_norm": 0.10451692716618664,
+      "learning_rate": 6.052139958933105e-06,
+      "loss": 0.4534,
+      "step": 479
+    },
+    {
+      "epoch": 4.203925845147219,
+      "grad_norm": 0.1014012869102775,
+      "learning_rate": 5.923223646466923e-06,
+      "loss": 0.4466,
+      "step": 480
+    },
+    {
+      "epoch": 4.212649945474373,
+      "grad_norm": 0.09600619629755003,
+      "learning_rate": 5.795585308423177e-06,
+      "loss": 0.4526,
+      "step": 481
+    },
+    {
+      "epoch": 4.221374045801527,
+      "grad_norm": 0.09629750961088542,
+      "learning_rate": 5.66922973159655e-06,
+      "loss": 0.4606,
+      "step": 482
+    },
+    {
+      "epoch": 4.230098146128681,
+      "grad_norm": 0.09815302673758064,
+      "learning_rate": 5.5441616546745646e-06,
+      "loss": 0.4541,
+      "step": 483
+    },
+    {
+      "epoch": 4.238822246455834,
+      "grad_norm": 0.0957082845975311,
+      "learning_rate": 5.420385768059908e-06,
+      "loss": 0.4572,
+      "step": 484
+    },
+    {
+      "epoch": 4.247546346782988,
+      "grad_norm": 0.10203231321540202,
+      "learning_rate": 5.297906713694523e-06,
+      "loss": 0.4576,
+      "step": 485
+    },
+    {
+      "epoch": 4.256270447110142,
+      "grad_norm": 0.0961706416633482,
+      "learning_rate": 5.176729084885508e-06,
+      "loss": 0.4572,
+      "step": 486
+    },
+    {
+      "epoch": 4.264994547437295,
+      "grad_norm": 0.10229924478117247,
+      "learning_rate": 5.0568574261328615e-06,
+      "loss": 0.4443,
+      "step": 487
+    },
+    {
+      "epoch": 4.273718647764449,
+      "grad_norm": 0.09666025616553983,
+      "learning_rate": 4.938296232959059e-06,
+      "loss": 0.4562,
+      "step": 488
+    },
+    {
+      "epoch": 4.282442748091603,
+      "grad_norm": 0.09933401963924726,
+      "learning_rate": 4.821049951740442e-06,
+      "loss": 0.457,
+      "step": 489
+    },
+    {
+      "epoch": 4.291166848418757,
+      "grad_norm": 0.095709128067926,
+      "learning_rate": 4.705122979540484e-06,
+      "loss": 0.4576,
+      "step": 490
+    },
+    {
+      "epoch": 4.299890948745911,
+      "grad_norm": 0.0945133866967145,
+      "learning_rate": 4.5905196639448816e-06,
+      "loss": 0.46,
+      "step": 491
+    },
+    {
+      "epoch": 4.308615049073064,
+      "grad_norm": 0.09068863647521308,
+      "learning_rate": 4.4772443028985004e-06,
+      "loss": 0.4493,
+      "step": 492
+    },
+    {
+      "epoch": 4.317339149400218,
+      "grad_norm": 0.09076000058999083,
+      "learning_rate": 4.3653011445442e-06,
+      "loss": 0.4578,
+      "step": 493
+    },
+    {
+      "epoch": 4.326063249727372,
+      "grad_norm": 0.09042479413132748,
+      "learning_rate": 4.254694387063514e-06,
+      "loss": 0.4599,
+      "step": 494
+    },
+    {
+      "epoch": 4.334787350054525,
+      "grad_norm": 0.0933053140254804,
+      "learning_rate": 4.1454281785191995e-06,
+      "loss": 0.458,
+      "step": 495
+    },
+    {
+      "epoch": 4.34351145038168,
+      "grad_norm": 0.08998117095396276,
+      "learning_rate": 4.037506616699682e-06,
+      "loss": 0.4564,
+      "step": 496
+    },
+    {
+      "epoch": 4.3522355507088335,
+      "grad_norm": 0.08810527149691408,
+      "learning_rate": 3.930933748965378e-06,
+      "loss": 0.4567,
+      "step": 497
+    },
+    {
+      "epoch": 4.360959651035987,
+      "grad_norm": 0.08798587461734628,
+      "learning_rate": 3.825713572096903e-06,
+      "loss": 0.4468,
+      "step": 498
+    },
+    {
+      "epoch": 4.369683751363141,
+      "grad_norm": 0.0908070109140394,
+      "learning_rate": 3.7218500321451754e-06,
+      "loss": 0.4527,
+      "step": 499
+    },
+    {
+      "epoch": 4.3784078516902945,
+      "grad_norm": 0.08670572701166608,
+      "learning_rate": 3.6193470242834415e-06,
+      "loss": 0.4549,
+      "step": 500
+    },
+    {
+      "epoch": 4.387131952017448,
+      "grad_norm": 0.0888975622499807,
+      "learning_rate": 3.518208392661184e-06,
+      "loss": 0.4583,
+      "step": 501
+    },
+    {
+      "epoch": 4.395856052344602,
+      "grad_norm": 0.08960109686027776,
+      "learning_rate": 3.4184379302599636e-06,
+      "loss": 0.4564,
+      "step": 502
+    },
+    {
+      "epoch": 4.404580152671755,
+      "grad_norm": 0.08892461210018777,
+      "learning_rate": 3.32003937875117e-06,
+      "loss": 0.4596,
+      "step": 503
+    },
+    {
+      "epoch": 4.413304252998909,
+      "grad_norm": 0.08863016698433596,
+      "learning_rate": 3.2230164283556918e-06,
+      "loss": 0.4511,
+      "step": 504
+    },
+    {
+      "epoch": 4.422028353326064,
+      "grad_norm": 0.08843132540215326,
+      "learning_rate": 3.1273727177055348e-06,
+      "loss": 0.4542,
+      "step": 505
+    },
+    {
+      "epoch": 4.430752453653217,
+      "grad_norm": 0.08957856998125675,
+      "learning_rate": 3.0331118337073494e-06,
+      "loss": 0.4665,
+      "step": 506
+    },
+    {
+      "epoch": 4.439476553980371,
+      "grad_norm": 0.08664626546623841,
+      "learning_rate": 2.9402373114079295e-06,
+      "loss": 0.457,
+      "step": 507
+    },
+    {
+      "epoch": 4.4482006543075245,
+      "grad_norm": 0.08685277411389214,
+      "learning_rate": 2.8487526338616224e-06,
+      "loss": 0.4538,
+      "step": 508
+    },
+    {
+      "epoch": 4.456924754634678,
+      "grad_norm": 0.0873940123339122,
+      "learning_rate": 2.7586612319996997e-06,
+      "loss": 0.4584,
+      "step": 509
+    },
+    {
+      "epoch": 4.465648854961832,
+      "grad_norm": 0.08665257770626869,
+      "learning_rate": 2.669966484501716e-06,
+      "loss": 0.4551,
+      "step": 510
+    },
+    {
+      "epoch": 4.4743729552889855,
+      "grad_norm": 0.08601547711556203,
+      "learning_rate": 2.582671717668772e-06,
+      "loss": 0.4658,
+      "step": 511
+    },
+    {
+      "epoch": 4.48309705561614,
+      "grad_norm": 0.09161088779549768,
+      "learning_rate": 2.4967802052987723e-06,
+      "loss": 0.4624,
+      "step": 512
+    },
+    {
+      "epoch": 4.491821155943294,
+      "grad_norm": 0.08958626962295069,
+      "learning_rate": 2.4122951685636674e-06,
+      "loss": 0.4644,
+      "step": 513
+    },
+    {
+      "epoch": 4.500545256270447,
+      "grad_norm": 0.08835234892005794,
+      "learning_rate": 2.3292197758886247e-06,
+      "loss": 0.4612,
+      "step": 514
+    },
+    {
+      "epoch": 4.509269356597601,
+      "grad_norm": 0.08502794085243678,
+      "learning_rate": 2.2475571428332276e-06,
+      "loss": 0.4472,
+      "step": 515
+    },
+    {
+      "epoch": 4.517993456924755,
+      "grad_norm": 0.08539786731552393,
+      "learning_rate": 2.1673103319746146e-06,
+      "loss": 0.4601,
+      "step": 516
+    },
+    {
+      "epoch": 4.526717557251908,
+      "grad_norm": 0.08660152357575181,
+      "learning_rate": 2.0884823527926377e-06,
+      "loss": 0.4574,
+      "step": 517
+    },
+    {
+      "epoch": 4.535441657579062,
+      "grad_norm": 0.08807963430843459,
+      "learning_rate": 2.0110761615569842e-06,
+      "loss": 0.4496,
+      "step": 518
+    },
+    {
+      "epoch": 4.544165757906216,
+      "grad_norm": 0.08713391947240909,
+      "learning_rate": 1.93509466121633e-06,
+      "loss": 0.4555,
+      "step": 519
+    },
+    {
+      "epoch": 4.552889858233369,
+      "grad_norm": 0.08887833508711152,
+      "learning_rate": 1.8605407012894438e-06,
+      "loss": 0.4543,
+      "step": 520
+    },
+    {
+      "epoch": 4.561613958560524,
+      "grad_norm": 0.0831347583772196,
+      "learning_rate": 1.7874170777583444e-06,
+      "loss": 0.4614,
+      "step": 521
+    },
+    {
+      "epoch": 4.570338058887677,
+      "grad_norm": 0.083557601522702,
+      "learning_rate": 1.7157265329634354e-06,
+      "loss": 0.4508,
+      "step": 522
+    },
+    {
+      "epoch": 4.579062159214831,
+      "grad_norm": 0.08536454712728841,
+      "learning_rate": 1.645471755500654e-06,
+      "loss": 0.4557,
+      "step": 523
+    },
+    {
+      "epoch": 4.587786259541985,
+      "grad_norm": 0.08426410904484165,
+      "learning_rate": 1.5766553801206575e-06,
+      "loss": 0.4509,
+      "step": 524
+    },
+    {
+      "epoch": 4.596510359869138,
+      "grad_norm": 0.08569779471913716,
+      "learning_rate": 1.5092799876299835e-06,
+      "loss": 0.456,
+      "step": 525
+    },
+    {
+      "epoch": 4.605234460196292,
+      "grad_norm": 0.08382181374993215,
+      "learning_rate": 1.4433481047943043e-06,
+      "loss": 0.4469,
+      "step": 526
+    },
+    {
+      "epoch": 4.613958560523446,
+      "grad_norm": 0.08306171541329403,
+      "learning_rate": 1.378862204243623e-06,
+      "loss": 0.4541,
+      "step": 527
+    },
+    {
+      "epoch": 4.6226826608506,
+      "grad_norm": 0.08568231225027859,
+      "learning_rate": 1.3158247043795735e-06,
+      "loss": 0.4593,
+      "step": 528
+    },
+    {
+      "epoch": 4.631406761177754,
+      "grad_norm": 0.08264784357921524,
+      "learning_rate": 1.2542379692847085e-06,
+      "loss": 0.4541,
+      "step": 529
+    },
+    {
+      "epoch": 4.6401308615049075,
+      "grad_norm": 0.0872470483995804,
+      "learning_rate": 1.1941043086338478e-06,
+      "loss": 0.4536,
+      "step": 530
+    },
+    {
+      "epoch": 4.648854961832061,
+      "grad_norm": 0.08482280818636626,
+      "learning_rate": 1.1354259776074472e-06,
+      "loss": 0.4474,
+      "step": 531
+    },
+    {
+      "epoch": 4.657579062159215,
+      "grad_norm": 0.08980570300270267,
+      "learning_rate": 1.0782051768070477e-06,
+      "loss": 0.4594,
+      "step": 532
+    },
+    {
+      "epoch": 4.6663031624863684,
+      "grad_norm": 0.08408980220342298,
+      "learning_rate": 1.0224440521727152e-06,
+      "loss": 0.4509,
+      "step": 533
+    },
+    {
+      "epoch": 4.675027262813522,
+      "grad_norm": 0.0848413070863153,
+      "learning_rate": 9.681446949025752e-07,
+      "loss": 0.4527,
+      "step": 534
+    },
+    {
+      "epoch": 4.683751363140676,
+      "grad_norm": 0.08329891519335332,
+      "learning_rate": 9.153091413744053e-07,
+      "loss": 0.4524,
+      "step": 535
+    },
+    {
+      "epoch": 4.692475463467829,
+      "grad_norm": 0.0823659135321468,
+      "learning_rate": 8.639393730692336e-07,
+      "loss": 0.4476,
+      "step": 536
+    },
+    {
+      "epoch": 4.701199563794984,
+      "grad_norm": 0.08462469191400637,
+      "learning_rate": 8.140373164970428e-07,
+      "loss": 0.4583,
+      "step": 537
+    },
+    {
+      "epoch": 4.709923664122138,
+      "grad_norm": 0.08587251152282088,
+      "learning_rate": 7.656048431245344e-07,
+      "loss": 0.4617,
+      "step": 538
+    },
+    {
+      "epoch": 4.718647764449291,
+      "grad_norm": 0.08534192176464622,
+      "learning_rate": 7.186437693049142e-07,
+      "loss": 0.4599,
+      "step": 539
+    },
+    {
+      "epoch": 4.727371864776445,
+      "grad_norm": 0.08224763625326706,
+      "learning_rate": 6.731558562097995e-07,
+      "loss": 0.4531,
+      "step": 540
+    },
+    {
+      "epoch": 4.7360959651035985,
+      "grad_norm": 0.08524884294020557,
+      "learning_rate": 6.291428097631524e-07,
+      "loss": 0.4522,
+      "step": 541
+    },
+    {
+      "epoch": 4.744820065430752,
+      "grad_norm": 0.08260361610070371,
+      "learning_rate": 5.866062805773176e-07,
+      "loss": 0.4547,
+      "step": 542
+    },
+    {
+      "epoch": 4.753544165757906,
+      "grad_norm": 0.08573726585911241,
+      "learning_rate": 5.455478638911071e-07,
+      "loss": 0.4553,
+      "step": 543
+    },
+    {
+      "epoch": 4.76226826608506,
+      "grad_norm": 0.0832532936654814,
+      "learning_rate": 5.05969099509982e-07,
+      "loss": 0.4551,
+      "step": 544
+    },
+    {
+      "epoch": 4.770992366412214,
+      "grad_norm": 0.08461747175902544,
+      "learning_rate": 4.6787147174831127e-07,
+      "loss": 0.4578,
+      "step": 545
+    },
+    {
+      "epoch": 4.779716466739368,
+      "grad_norm": 0.08296914778867019,
+      "learning_rate": 4.3125640937368373e-07,
+      "loss": 0.4638,
+      "step": 546
+    },
+    {
+      "epoch": 4.788440567066521,
+      "grad_norm": 0.0838735021411658,
+      "learning_rate": 3.961252855533504e-07,
+      "loss": 0.4537,
+      "step": 547
+    },
+    {
+      "epoch": 4.797164667393675,
+      "grad_norm": 0.08335291392244688,
+      "learning_rate": 3.6247941780271026e-07,
+      "loss": 0.4556,
+      "step": 548
+    },
+    {
+      "epoch": 4.805888767720829,
+      "grad_norm": 0.08399000470045154,
+      "learning_rate": 3.3032006793590977e-07,
+      "loss": 0.4622,
+      "step": 549
+    },
+    {
+      "epoch": 4.814612868047982,
+      "grad_norm": 0.08546470506503838,
+      "learning_rate": 2.9964844201851193e-07,
+      "loss": 0.4516,
+      "step": 550
+    },
+    {
+      "epoch": 4.823336968375136,
+      "grad_norm": 0.08980523198570832,
+      "learning_rate": 2.704656903222791e-07,
+      "loss": 0.4645,
+      "step": 551
+    },
+    {
+      "epoch": 4.8320610687022905,
+      "grad_norm": 0.08327924025302387,
+      "learning_rate": 2.4277290728202063e-07,
+      "loss": 0.4559,
+      "step": 552
+    },
+    {
+      "epoch": 4.840785169029444,
+      "grad_norm": 0.08394210027215841,
+      "learning_rate": 2.1657113145455045e-07,
+      "loss": 0.4518,
+      "step": 553
+    },
+    {
+      "epoch": 4.849509269356598,
+      "grad_norm": 0.08226803659766549,
+      "learning_rate": 1.9186134547974912e-07,
+      "loss": 0.4386,
+      "step": 554
+    },
+    {
+      "epoch": 4.858233369683751,
+      "grad_norm": 0.08318642805856608,
+      "learning_rate": 1.6864447604370004e-07,
+      "loss": 0.4536,
+      "step": 555
+    },
+    {
+      "epoch": 4.866957470010905,
+      "grad_norm": 0.08311610244579633,
+      "learning_rate": 1.46921393843944e-07,
+      "loss": 0.4655,
+      "step": 556
+    },
+    {
+      "epoch": 4.875681570338059,
+      "grad_norm": 0.08447920109375157,
+      "learning_rate": 1.2669291355681623e-07,
+      "loss": 0.4504,
+      "step": 557
+    },
+    {
+      "epoch": 4.884405670665212,
+      "grad_norm": 0.08099018291123877,
+      "learning_rate": 1.0795979380690657e-07,
+      "loss": 0.4492,
+      "step": 558
+    },
+    {
+      "epoch": 4.893129770992367,
+      "grad_norm": 0.08134813316578475,
+      "learning_rate": 9.072273713860213e-08,
+      "loss": 0.4591,
+      "step": 559
+    },
+    {
+      "epoch": 4.9018538713195206,
+      "grad_norm": 0.08260979942319405,
+      "learning_rate": 7.498238998973506e-08,
+      "loss": 0.4524,
+      "step": 560
+    },
+    {
+      "epoch": 4.910577971646674,
+      "grad_norm": 0.08402615351585943,
+      "learning_rate": 6.073934266735303e-08,
+      "loss": 0.4479,
+      "step": 561
+    },
+    {
+      "epoch": 4.919302071973828,
+      "grad_norm": 0.08162169304255132,
+      "learning_rate": 4.799412932555925e-08,
+      "loss": 0.4523,
+      "step": 562
+    },
+    {
+      "epoch": 4.9280261723009815,
+      "grad_norm": 0.08091395918593658,
+      "learning_rate": 3.6747227945506113e-08,
+      "loss": 0.4612,
+      "step": 563
+    },
+    {
+      "epoch": 4.936750272628135,
+      "grad_norm": 0.08162506604523725,
+      "learning_rate": 2.699906031745414e-08,
+      "loss": 0.4604,
+      "step": 564
+    },
+    {
+      "epoch": 4.945474372955289,
+      "grad_norm": 0.08577808539743104,
+      "learning_rate": 1.874999202494454e-08,
+      "loss": 0.4628,
+      "step": 565
+    },
+    {
+      "epoch": 4.9541984732824424,
+      "grad_norm": 0.08253647896585553,
+      "learning_rate": 1.2000332431121309e-08,
+      "loss": 0.4551,
+      "step": 566
+    },
+    {
+      "epoch": 4.962922573609596,
+      "grad_norm": 0.08410829706038782,
+      "learning_rate": 6.750334667091629e-09,
+      "loss": 0.4557,
+      "step": 567
+    },
+    {
+      "epoch": 4.971646673936751,
+      "grad_norm": 0.08212542706662271,
+      "learning_rate": 3.000195622453461e-09,
+      "loss": 0.458,
+      "step": 568
+    },
+    {
+      "epoch": 4.980370774263904,
+      "grad_norm": 0.08315321166610817,
+      "learning_rate": 7.500559379103323e-10,
+      "loss": 0.4625,
+      "step": 569
+    },
+    {
+      "epoch": 4.989094874591058,
+      "grad_norm": 0.08310709809675666,
+      "learning_rate": 0.0,
+      "loss": 0.4504,
+      "step": 570
+    },
+    {
+      "epoch": 4.989094874591058,
+      "step": 570,
+      "total_flos": 1.517527475894721e+19,
+      "train_loss": 0.18812398983721149,
+      "train_runtime": 53023.3041,
+      "train_samples_per_second": 5.533,
+      "train_steps_per_second": 0.011
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 570,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.517527475894721e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}